❌

Normal view

iPhone Dystopia: Age Verification Baked Into IOS 27.2 Using Camera

29 October 2026 at 14:12
It’s just one step closer to dystopia as Apple caves to government pressure for age verification. It lets Apple play in major markets around the world: Australia, Brazil, England, and South Korea. But it will also be used in the United States. Facebook uses a selfie to prove age to get a “verified age” badge, but that is an app; Apple baked it into the operating system.

Here are Apple’s claims:

  • The selfie “isn’t accessible to anyone, not even Apple.”
  • “Once your age is confirmed, your selfie is deleted and isn’t retained in any form.”
  • Apple says PCC data is encrypted in transit and deleted afterward.

However, a user’s picture is sent to Apple’s cloud servers to be processed, where the data points remain. The picture is not needed at that point and can be deleted.

Governments can now mandate that Apple turn the feature on regardless of which apps are installed on the phone. Secondarily, governments can mandate that the apps access the feature before granting to access to their services. ⁃ Patrick Wood, Editor.

When Apple released iOS 27.2 on Sunday, it crossed a threshold no major operating system had crossed before: embedding a government-compelled biometric age gate directly into the OS layer itself — not inside an app, not inside a browser, but in the platform that every app runs on top of. For the first time, accessing adult-rated content on an iPhone or iPad in Australia, Brazil, Singapore, or the United Kingdom now requires passing a facial age estimation scan administered by the operating system before the App Store will proceed.

South Korea is also part of the five-country rollout, but Apple’s selfie-based facial scan is not available there — South Korean users face annual carrier-verified identity checks instead. Adults who fail or decline the facial scan do not get a different app or a competing service; they get a permanently restricted device — or they get a different phone.

That architectural choice — OS-level rather than app-level — is the fact that makes iOS 27.2’s biometric gate different from every prior age-verification rollout in the industry, and it is the fact Apple’s public documentation downplays. Once this plumbing is baked into the operating system, every future government mandate that wants to require it will find Apple’s infrastructure already installed. That is not a side effect. It is the point.

How Apple’s Four-Angle Face Scan Works

Apple’s own support documentation describes the process as a liveness check more than an identity interrogation. Users are instructed to perform a deliberate four-angle head rotation — right, left, up, down — capturing their face from multiple perspectives. Apple advises removing anything that might obscure a result: masks, hats, sunglasses. The company calls the underlying technology an “age-estimation model,” not a facial recognition or identity-matching system, which is a meaningful distinction: estimation predicts a probabilistic age range from facial geometry rather than matching the face to a database record.

The models run “both on-device and on Private Cloud Compute,” Apple says — the hybrid on-device/cloud framework it originally launched for Apple Intelligence AI features. Results are “combined on your device.” Apple’s PCC architecture is designed so that sensitive data exchanged with cloud services is encrypted and deleted afterward, with the process available for researcher verification. The company has since expanded PCC to Google Cloud, though it says its core security model remains unchanged.

Apple promises the selfie “isn’t accessible to anyone, not even Apple,” and that “once your age is confirmed, your selfie is deleted and isn’t retained in any form” per its selfie verification support page. Users who are not cleared as adults are directed to try again with a fresh selfie, or to use an alternative: a credit card on file, Apple ID account-history signals, or a government-issued ID scan.

What Apple Has Not Said, and Why It Matters

Apple has not disclosed how accurate its age-estimation model is, what its misclassification rates look like across age groups or demographic categories, or what data record — if any — is retained after a scan is processed and before deletion is confirmed.

The Computer & Communications Industry Association (CCIA) has cited the most authoritative public benchmark on this question: NIST’s Face Analysis Technology Evaluation (FATE). Across tested methodologies, the CCIA notes that one-year age estimation accuracy for 13-year-olds ranged from just 7.2% to 34.5%. Being estimated one year too young can cause a legal adult to lose access to all 18+ apps without parental consent — a mismatch that cuts in both directions and scales with the size of the affected population.

Apple has not opened its PCC age-estimation pipeline to independent audit in the way it has with other PCC workloads. Its PCC SOC 3 audit — while validating general controls — explicitly states that “the independent auditor does not express an opinion or any other form of assurance about Apple’s artificial intelligence services.” This means that NIST FATE’s findings on comparable systems are the best available public proxy for Apple’s likely accuracy — but they are not Apple’s actual numbers, and Apple has provided no mechanism for anyone outside the company to determine whether its model performs better or worse than the NIST benchmark range.

The CCIA has also noted that approximately 11% of U.S. adults lack government-issued photo ID, meaning the government ID fallback option is not universally available to adults who fail the facial scan and cannot produce a credit card or a qualifying account history. At OS level, there is no third exit — no different app, no competing service, no platform workaround.

Regulatory Patchwork That Forced Apple’s Hand

The five-country rollout reflects a complex global patchwork of laws that converged on Apple’s timeline.

Texas. The most immediate domestic driver is Texas’s App Store Accountability Act (SB 2420), which moved the age gate from individual websites to the account required to install anything. New Apple Accounts created in Texas are subject to the law from June 4, 2026, requiring age assurance and parental consent for minors under 18 for downloads, in-app purchases, and significant changes to app settings. Under-18 users must be in a Family Sharing group, with Ask to Buy and Screen Time features automatically enabled and undisable by parents. For all accounts, age range is shared with developers by default, with no opt-out available.

Apple was not a willing participant. The company previously fought the law, describing it as requiring collection of “sensitive personal identifying information” for every Texan who wants to download an app. Tim Cook, then CEO, personally called Governor Greg Abbott urging a veto. A federal judge blocked the law in December 2025 on First Amendment grounds. The Fifth Circuit lifted that block in June 2026. On July 6, 2026, the Supreme Court declined to block enforcement while the underlying case continues.

Australia. Six age verification codes came into effect on March 9, 2026, requiring proof of age to access pornography, R18+ games, adult apps, and explicit AI chatbots. The codes apply to app stores — including Apple’s App Store and Google Play — under penalties of up to AUD $49.5 million per breach. Apple announced Australia, Brazil, Singapore age checks on February 24, 2026, blocking downloads of 18+ apps unless users were confirmed adults. The full app store age assurance enforcement deadline fell on September 9, 2026 — six weeks before iOS 27.2 arrived.

United Kingdom. Under the Online Safety Act, Ofcom defines “highly effective” age assurance to explicitly include facial age estimation using biometric technology. Ofcom has already moved from guidance to enforcement action, fining 8579 LLC £1.35 million for failing to implement legally required age verification checks. Apple launched UK age verification tools in March 2026 ahead of stricter enforcement requirements.

More States Coming; Tim Cook’s Public Posture Has Shifted

The U.S. state-by-state expansion shows no sign of slowing. Louisiana’s version of the app store accountability law becomes effective July 1, 2027, while Alabama’s version goes further, requiring every existing account in the state to be age-verified by October 1, 2027.

The shift in Tim Cook’s public posture — he is now Apple’s Executive Chairman after handing the CEO role to John Ternus in September 2026 — is striking against the backdrop of Apple’s earlier legal resistance. Cook has welcomed the EU KIDS Act, saying: “We’re very encouraged that the EU Kids Act proposed by the (European) Commission has helped trigger a constructive conversation around this critical issue in Europe.” The EU KIDS Act explicitly puts app stores and operating systems within its scope — meaning the biometric pipeline Apple just built for five countries would serve European mandates without additional infrastructure development. (Separately, Apple has said Siri AI will not ship on iOS 27 in the EU at launch due to the Digital Markets Act — a reminder that the EU’s regulatory posture toward Apple remains complex even as Cook signals openness on age verification.)

Cook lobbied personally against age verification in Texas. He publicly welcomes it in Europe. The difference between those two positions is not a change in Apple’s values; it is a change in Apple’s compliance posture once the infrastructure is built and the legal battles are lost. The age gate is now a product feature.

An Industry Pattern and a Structural Concern

Apple is not alone in deploying facial analysis for identity purposes. Meta now asks Facebook users for a video selfie to obtain a “verified” badge, without publicly disclosing how long it retains those videos. Google’s model in Canada sends those it estimates as minors to verify their age with an ID or a selfie. Reddit’s age checks in Europe go through Persona, which, according to researchers, runs up to 269 verification checks on each user.

What distinguishes Apple’s implementation is architectural: not inside a single app or website, but embedded in the operating system itself. A user who dislikes Meta’s video selfie requirement can stop using Facebook. A user who dislikes Apple’s iOS 27.2 age scan cannot stop using the App Store without giving up iOS. The failure mode at this level is not correctable by switching apps; it is correctable only by switching platforms.

The CCIA has previously noted that public support for age checks in the abstract evaporates when people actually face uploading IDs or submitting to face scans. At OS level, individual resistance is harder to exercise. Users cannot avoid a single website or delete an app; they must confront the gate any time they try to install anything in an 18+ category, or they must permanently surrender the ability to make that choice.

Once Baked In, the Plumbing Stays

Apple’s promise that the selfie is deleted after processing may or may not be independently verifiable — the company has not opened its PCC age-estimation pipeline to the kind of independent security review it has extended to other PCC workloads. Whether the selfie is retained is, in some ways, less important than what the scan normalizes: a biometric facial scan as a routine condition of using a device.

Privacy researchers have noted that this normalization matters beyond the specific data question. Every jurisdiction that has been considering an age verification mandate — and there are now dozens — will observe that Apple built this infrastructure, deployed it smoothly across five countries, and complied with regulators across three continents simultaneously. Louisiana and Alabama are already on the calendar. The EU KIDS Act is welcomed by Apple’s own executive chairman. The global trajectory is not ambiguous.

The question is no longer whether operating systems will carry mandatory biometric age gates. iOS 27.2 has answered that. The questions are: how accurate will those gates be, who audits them, what happens to adults who are misclassified, and what other mandates will the same infrastructure eventually serve — because the plumbing Apple installed Sunday does not have a single-use expiration date.

Currency conversions in this article are approximate and based on rates current at time of publication.

Read full story here…

Stablecoin Transfers Quadruple To $891B As US Banks Expand Adoption

28 October 2026 at 14:33
It’s sweeping the world. Stablecoins are the privatized version of CBDCs, or Central Bank Digital Currencies. The Trump family is a major player in the space with World Liberty Financial. The banks fought it tooth and nail at the start, but now they are caving in. The flip side of stablecoins is asset-backed tokens, which are scaling even faster than stablecoins.

I have warned about this in my latest three books, written in the last 12 months. The pinnacle of these, Crucible of Tokens, is required reading if you want to get ahead of the tsunami.

Every token that touches a physical asset – real estate, stocks & bonds, Treasuries, mutual funds – takes the title out of your hands. For the unsuspecting, this ends private property. In 2026 alone, the stock exchanges are being tokenized. Believe me, this will be the biggest heist in the history of the world.⁃ Patrick Wood, Editor.

Coinbase reports growing adoption by banks and businesses, even as stablecoin market capitalization remains largely unchanged over the past year.

Stablecoins moved more than $400 million an hour in adjusted transfer volume during the third quarter of 2026, as quarterly activity reached $891 billion, more than quadrupling since early 2023, according to a new report commissioned by Coinbase.

The report found that 73% of screened stablecoin transfer volume occurred outside US banking hours, including nights, weekends and holidays.

Distribution of stablecoin transfer volume by time of day. Source: Coinbase and The Block Research, via Dune Analytics

Seven of the 10 largest US banks have tested stablecoins or tokenized deposits for settlement, while 30% of mid-market companies conducting cross-border business are using or piloting stablecoins, up from 10% in 2024. Business-to-business payment volume also rose 36% year over year.

Conducted by The Block Research, the findings identified 293 million blockchain addresses holding stablecoins, exceeding the combined number of cards issued by American Express and Discover, though researchers cautioned that addresses do not necessarily represent individual users.

The report also estimated that moving 80% of international transactions onto stablecoin networks could add $155 billion to global GDP annually and free up roughly $3 trillion in liquidity currently held by banks to support cross-border payments.

Those projections are based on hypothetical adoption scenarios rather than observed economic gains. Researchers also acknowledged that widespread stablecoin adoption could reduce bank lending, potentially offsetting some of the benefits of faster and cheaper payments.

A separate survey released Monday by the National Cryptocurrency Association found that 42% of US adults, representing approximately 112 million people, would be more likely to open an account with a bank offering crypto services or use that institution for more of their financial needs.

The survey found that 18% of US adults would like to buy or use stablecoins through their primary bank or financial institution. However, 64% said their bank, credit union or financial advisor had never raised the topic of cryptocurrency, while 43% did not know whether their bank offered crypto services.

Americans’ interest in crypto banking services by demographic group.
Source: National Cryptocurrency Association, 2026 Crypto at the Bank Report

Read full story here…

Blessing Or Curse: Pax Silica In The Philippines

Pax Silica is about building empire, exporting the means of Technocratic control and resource extraction from host nations. Despite pushback over minor issues, the leader of the Philippines, Ferdinand Marcos, has widely praised the program, assuring that “no foreign power can exploit local minerals or compromise Philippine territory.” This brings to mind the pirates’ creed of rape, pillage, and plunder. ⁃ Patrick Wood, Editor.

AS the world sees the rising influence of artificial intelligence (AI), the Philippines was presented with the Pax Silica deal that could position the country as an AI innovation hub in Southeast Asia.

While supporters view the project as an economic catalyst, its critics argue that the massive initiative poses a threat to national interest and local communities.

Ambitious endeavor The United States came up with the idea of Pax Silica in Dec. 2025 to counter China’s dominance in technology manufacturing. It aims to secure supply chains for artificial intelligence, semiconductors, and critical minerals.

China is said to be controlling about 90% of rare earth refining that allows for the manufacture of advanced chips. But in 2025, the country implemented rare earth controls in response to US tariffs.

The proposed Pax Silica hub in the Philippines, envisioned as the Luzon Economic Corridor, covers a total area of 1,620 hectares (approximately 4,000 acres) in New Clark City in Capas, Tarlac, a former US military base.

Original members of the Pax Silica alliance include the United States, Japan, South Korea, Singapore, the United Kingdom, Israel, the United Arab Emirates, and Australia. Now it has over 30 member states. The Philippines joined the initiative in April.

Trade Undersecretary Ceferino Rodolfo said the Philippine government will be signing a framework deal with a US government agency for Pax Silica bound by Philippine laws.

Bases Conversion and Development Authority (BCDA) President and CEO Joshua Bingcang earlier said that the three-phase project is expected to start in 2028.

Good deal? President Ferdinand Marcos Jr. gave his full support to Pax Silica, describing it in his fifth State of the Nation Address as a center in the global AI and technology value chain. “It will bring quality jobs to our people, accelerate our industrial competitiveness, and revitalize our economy.” According to Bingcang, Pax Silica can pull in investments of up to $30 billion to $70 billion and could generate over 130,000 jobs for Filipino engineers, researchers, computer science graduates, and other highly skilled professionals.

Other government agencies that expressed support for the initiative are the Office of the President, Department of Trade and Industry, BCDA, and the National Water Resources Board.

Political analyst Dennis Coronacion said the Pax Silica project would definitely be identified as one of Marcos’ most ambitious projects even if its operations begin past his term.

Cost But critics warned that the massive technological infrastructure would demand a costly toll.

AI data centers and semiconductor facilities require massive daily cooling volumes estimated at 65 million to 90 million liters of water per day, raising questions as to whether the area can sustain the project without affecting existing supplies.

Also, the high-density energy needs of servers and industrial processing threaten to overburden the local power grid and drive up regional electricity rates.

The footprint targets roughly 1,600 to over 20,000 hectares for industrial lots and supporting renewable energy/reservoir infrastructure, risking local wetlands, watersheds, and agricultural lands.

Environmental and indigenous groups have likewise raised alarms over the potential displacement of thousands of farmers and local Aeta communities.

Because of this, at least 70 organizations have voiced their opposition to the project.

The Momentum Research survey done in August showed that 34 percent of Filipinos were against the project, 26 percent admitted having little to no knowledge of it, while 20 percent say it needs further studies. Whereas, only eight percent are supportive and 11 percent are totally unaware of the planned AI hub.

“There is no technology sharing involved, which is what we need right now instead of just playing as the host of these tech projects,” Dennis Coronacion, an associate professor of Political Science at the University of Santo Tomas, told The Manila Times.

“When they left, we still were not high-tech,” he added.

In a Senate hearing, Bingcang said the project would not rely on groundwater extraction, which he said should be reserved for residential and agricultural use.

The BCDA, he said, was planning to tap surface water from nearby rivers and store it for treatment, using detention tanks to capture excess water, particularly during the rainy season.

Bingcang identified the Bangot River as the project’s closest potential water source, along with the O’Donnell and Abacan rivers.

He said the initial studies showed that the Bangot River alone could yield 130 million to 200 million liters of water.

“So it is far more than what is needed for this project,” he said in Filipino.

The initial assessment, Bingcang said, was conducted during the dry season and that participating industries would be required to recycle water.

“And based on the industry average, at least 70 percent is recycled. So they only have to top up a small amount for their requirements,” he said.

Desalination of seawater is also seen as an alternative water source.

To supply its estimated 3-gigawatt (3 million kilowatts) power requirement, the BCDA said Pax Silica aims to use clean and green energy, including solar farms with an option of battery storage.

On the issue of community displacement, Palace press officer Claire Castro said the government would ensure affected residents would not lose their livelihoods should relocation become necessary.

Coronacion said the entire project, including projected costs and proposed countermeasures, must all be subjected to thorough study.

“Before we agree to this plan, we should do a thorough study. We will see if our concerns about the environment and the displaced local farmers will be addressed. Until those issues are resolved, we should not say yes,” Coronacion said.

Geopolitics AI domination is practically a war between China and the US.

Whoever possesses the most advanced technology has the edge in the rivalry and likely wins, Coronacion said.

“This is the mindset of these dominant powers. They want to exclude their competitors in the AI ecosystem,” he said.

Another issue hounding Pax Silica is the possibility that it could grant foreign entities excessive authority over domestic land, resources, and operational governance.

In response, Marcos assured that no foreign power can exploit local minerals or compromise Philippine territory.

Read full story here…

California Startup Says It Has Injected AI-Designed Virus Into Humans For First Time

The company said, “This milestone marks an early step toward translating AI-guided genome design into personalized therapies.” The doors are flung wide open for using AI to modify life itself. Unintended consequences? That’s an afterthought. These lunatic Technocrat scientists will never stop short of being escorted out of their buildings, deactivating their ID cards, razing their buildings to the ground, and then banning them from ever experimenting with human life again. ⁃ Patrick Wood, Editor.
California startup Radical Numerics says it has administered an AI-designed bacteriophage, Omnii-1, to a human patient as a compassionate‑use treatment for an antibiotic‑resistant infection, marking a claimed first in AI‑generated phage therapy.

An artificial intelligence research lab in California said it has administered a custom AI-designed virus into a human patient for the first time as a potential scalable solution to counter antibiotic-resistant superbugs.

AI lab Radical Numerics announced the treatment on Tuesday, but it is pending scientific scrutiny in the form of a peer-reviewed study.

If verified, the treatment, named Omnii-1, could mark the first time AI-designed bacteriophages have been administered to a human patient.

These phages are viruses that specifically infect bacteria. Antibiotic-resistant superbugs are projected to cause nearly 40 million deaths by 2050.

Under current phage treatment approaches for bacterial infections, researchers undertake a time-consuming search of vast environmental collections of phages to find one that happens to match a patient’s specific infection.

It starts with clinicians extracting a sample of the bacteria causing a patient’s infection to find or design phages that kill the infectious bug.

Key to this challenge is understanding the genetic features of a potential phage that could kill particular bacterial strains.

Phages kill bacteria through a multi-step process in which the virus first binds to specific receptor molecules on the surface of the target bacteria.

They then proceed to enter the bacterial cell to hijack it for their own replication and break it open.

The initial binding of the virus to bacteria is very specific, as a given phage only targets particular strains or species of bacteria.

This specificity is a crucial advantage of phage therapy, which spares beneficial gut microbes that tend to be affected by broad-spectrum antibiotics.

Radical Numerics claims that their AI-driven approach can lead to rapid, on-demand generation of phages that meet patient requirements, accelerating the timeline for treatment.

“In the world’s first clinical use of AI-designed phages, designs generated by Omnii-1 were administered to a patient with an antibiotic-resistant bacterial infection under an authorised, physician-led compassionate-use pathway,” Radical Numerics wrote in a blog post.

“This milestone marks an early step toward translating AI-guided genome design into personalised therapies,” it said.

The startup claimed that its AI model is trained on diverse genomic sequences and phage genomes, as well as data on previously measured phage-host interactions.

Learning from this information, the AI could identify, as well as design, candidate phages for specific therapeutic requirements.

“We synthesise Omnii-1-designed phage variants and measure their recovery and activity across bacterial strains,” the startup said.

Recovery is a measure of a phage’s viability, which says whether an AI-designed viral genome can successfully come to life and replicate in the lab.

“These experiments help characterise the activity of our variants against target hosts and inform subsequent rounds of modelling and design,” Radical Numerics explained.

The company said its Omnii-1 designs were synthesised and administered to a human patient with an antibiotic-resistant infection who had very few remaining medical options.

But it did not reveal specific details of the treatment’s clinical outcomes, citing patient privacy.

Read full story here…

Brace Yourselves: Serious Trouble Brewing In The Data Center Industry

23 October 2026 at 16:24
Why has the AI industry been in such a rush to swamp the country with massive data centers while throwing caution to the wind? It’s the only way that they can be assured to get enough compute in place to run the entire technocratic system that they have planned: Tokenization of all financial assets and trading systems, digital ID and authentication, the AI takeover of government operations, ubiquitous social control, etc. This is the system that will survive the coming meltdown. ⁃ Patrick Wood, Editor.

Ever since the launch of ChatGPT in 2022, a torrent of money has flowed into building out data centers to power the AI boom.

The efforts reached a fever pitch the day after Donald Trump’s second inauguration in 2025, when a bevy of AI luminaries including OpenAI CEO Sam Altman and Softbank founder Masayoshi Son crowded into the Oval Office to announce a $500 billion private investment in data centers.

But as the public has grown jittery about the impacts of AI on society and the job market — and story after story has landed about the downsides of hosting data centers in local communities — the buildout has emerged as a major political flashpoint and a key issue in the looming midterms.

The data industry’s headwinds aren’t just political, either. As the Financial Times reports, serious cracks are forming at the financial and logistical level, especially as serious AI revenue remains elusive.

Numbers cited by the paper are eye-popping. Drawing on data from regulatory intelligence outfit RegLayer, the FT reports that 47 large data center projects have already fallen through this year. Analysts at the equity research firm Bernstein only believe that about 35 percent of planned facilities are feasible.

All those canceled projects are due to a lot of factors, including local communities mobilizing against them, but it’s also macroeconomic. The buildout is taking place during a period of high interest rates and shaky bond markets, and people involved with the effort told the paper that the situation is “threatening their ambitions.”

Some of the plans were shaky to begin with. The FT reports that top labs like Anthropic and OpenAI have pushed so hard for the developments, believing them to be key to their success, that they’ve swallowed outsized costs and deemphasized due diligence. Investors are taking notice: stock in CoreWeave, a cloud provider closely linked to the AI data center industry, has fallen by more than 27 percent over the past month.

The really big question, though, is what happens to all these data center projects if the bottom starts to fall out of the AI industry financially, a situation that many economists have warned is a possibility over the past few years. For now, things are still chugging along — albeit much more slowly than anticipated.

“People aren’t just dropping their shorts for data centers anymore,” one unnamed executive told the FT. “It’s much harder today to finance a data center.”

Read full story here…

Delivery Robots Accused Of Scanning Customers’ Faces, Sharing Biometric Data

Cute little delivery robots rolling down the sidewalk turn into a Chucky horror movie as they grab your face with a video scan and other information about you when you retrieve your lunch (or whatever you ordered). All of the innocent bystanders are all scanned, too. Is the microphone always on, picking up all conversations along the route? You don’t need to wait for this court case to settle: don’t use delivery robots in the first place. ⁃ Patrick Wood, Editor.

In a startling development that strikes at the heart of the burgeoning autonomous technology sector, several delivery robots are now facing serious allegations of scanning customers’ faces and harvesting biometric data. The accusations suggest that these automated machines, designed for convenience, may be functioning as mobile surveillance units, capturing highly sensitive personal information without explicit user consent.

The controversy centers on how these robots process visual data while navigating sidewalks and interacting with clients. While companies claim the cameras are strictly for navigation and obstacle avoidance, whistleblowers and privacy experts suggest that the software may be performing advanced facial mapping. This data, which is unique to every individual, could potentially be stored, analyzed, or even shared with third-party advertisers and data brokers.

Why This Matters

BozokMedia analysis shows that this incident marks a critical turning point in the debate over ‘surveillance capitalism.’ As autonomous robots become a common sight in urban environments, the line between helpful automation and intrusive monitoring becomes dangerously blurred. If biometric data is leaked or misused, the implications for individual security and identity theft are profound.

The integration of facial recognition into consumer-facing robotics without a robust legal framework is a ticking time bomb for digital privacy.

Legal experts are already examining whether current data protection laws, such as the GDPR in Europe or various state-level privacy acts in the US, are sufficient to handle the nuances of mobile robotic surveillance. The core issue lies in the ‘passive collection’ of data—where bystanders and customers are recorded simply by being in the vicinity of the robot.

Historical Background

The history of automated surveillance has evolved from stationary CCTV cameras to highly mobile, AI-driven devices. In previous decades, facial recognition was restricted to high-security government installations. Today, the rapid commercialization of AI has brought this technology into the streets, often outpacing the ability of lawmakers to regulate it effectively.

Did You Know?: Biometric data, such as facial geometry, is considered one of the most sensitive forms of personal information because, unlike a password, it cannot be changed if compromised.

As the industry moves forward, the pressure is mounting on tech giants to implement ‘privacy-by-design’ protocols. This would involve ensuring that visual data is processed locally on the device and immediately deleted, rather than being uploaded to a cloud server for long-term storage.

Read full story here…

3D-Printed Fake Meat and Fish Finally Hitting American Grocery Shelves

This is Technocrat’s dream: to make food for the masses in a sterile factory without any animal protein. It has been years in the making (research and development), but it’s finally arrived – 3D-printed meat, fish, and chicken. Of course, the label doesn’t say that, because who in their right mind would eat it? Can you picture your child saying, “Oh look, Mommy, can we have 3D-printed steak for dinner?” Well, at least it’s not “soylent green,” but it’s just as repulsive. ⁃ Patrick Wood, Editor.

3D-printed fake meat and fish have officially hit American grocery stores. Israeli food-technology company Steakholder Foods announced the U.S. retail launch of its PERFECTA brand, introducing 3D-printed meat and seafood to grocery stores across the Northeastern United States.

The packaging DOES NOT disclose that the company’s products are manufactured using 3D-printing technology. Instead, the products are marketed as “premium plant-based meat.”

Steakholder Foods is rolling out five products under its PERFECTA brand:

  • “Marbled Steak”
  • “Filet Mignon”
  • “Chicken Breast”
  • “Salmon Patties”
  • “White Fish Patties”

Its plant-based steak formulation contains soy flour, rapeseed oil, modified starch, methylcellulose, xanthan gum, locust bean gum, and artificial meat coloring designed to imitate the appearance of beef.

These are not ingredients I would put in my body.

One thing is for sure, I will NOT be purchasing this 3D-printed slop.

Read full story here…

💾

https://substack.com/@UCdbrACyGhhbaqJv6oAGJjJQ 3D-printed plant-based meat and fish products have officially entered U.S. grocery stores. Steakholder Foods,...

BlackRock’s Record Asset Holdings Are Half The Size Of US Economy

16 October 2026 at 16:34
BlackRock is headed by Larry Fink, a prominent member of the TRILATERAL COMMISSION and co-CEO of the World Economic Forum. He has autocratic control over $15.49 trillion in assets, almost half of the US economy. Fink says “we are at the beginning of the tokenization of all assets” and on a single blockchain. As an Arch-Technocrat, he makes Elon Musk look like a financial midget. ⁃ Patrick Wood, Editor.

BlackRock ended the third quarter with $15.49 trillion in assets under management, a record, after $205 billion of net client inflows, according to its earnings release filed with the SEC on Oct. 14.

GAAP earnings came to $10.57 a share, and as-adjusted earnings to $15.93. Revenue rose 19 percent to $7.75 billion, and the company said it has raised $68 billion in private markets over the last twelve months. Its AI Infrastructure Partnership made its first deployment during the quarter.

Chairman and CEO Larry Fink said clients “entrusted us with $205 billion of net inflows this quarter, and $869 billion over the last year.” He added: “We’re entering our seasonally strongest fourth quarter with accelerating momentum.”

As the Epoch Times notes further, the company drew almost $135 billion in net new flows into its flagship iShares exchange-traded funds, also known as ETFs.

It also repurchased $550 million of its shares during the quarter, delivering stronger value to current shareholders.

“Market fundamentals are strong and well supported, with higher margins and earnings momentum catalyzed by new technology,” CEO Larry Fink said in a statement.

Looking ahead to the final three months of 2026, Fink noted that BlackRock has plenty of momentum and a broad pipeline of opportunities across multiple business segments, including private markets and retirement funds.

BlackRock popped as much as 5 percent following the earnings data but erased its gains as the midweek session progressed.

Private Credit Improvements

The private credit industry has been under tremendous strain this year, reflecting the so-called “SaaSpocalypse” and increasing concerns surrounding lenders’ exposure to software companies.

This was observed in the wave of redemption requests.

But BlackRock’s push into private credit markets does not seem affected by the sector’s widespread pressures.

In a call with analysts, CFO Martin Small suggested that investment demand for private credit has been solid.

Net inflows into its private credit funds rose by $5 billion from the previous quarter, according to the earnings report.

“We think there’s a much larger opportunity that’s more durable in delivering private high-grade private credit in for debt across our platform and just deepening our client relationships over time,” Small said.

Additionally, BlackRock’s client redemptions, which have been in focus for the broader industry since the start of 2026, have eased modestly.

Redemption requests for its Corporate Lending Fund fell to about 11.5 percent of shares in the latest quarter, down from 13.3 percent in the second quarter.

But while private credit stress appears to be dissipating, Moody’s warned the situation could be worse than reported default rates suggest.

“Asset quality risks are mounting among weaker middle market borrowers in the US private credit market as higher-for-longer interest rates, refinancing challenges, redemption pressures and persistent concerns over transparency weigh on performance,” the firm said in an Oct. 14 report.

Earnings Season So Far

Third-quarter earnings season started on Oct. 13, beginning with some of the big banks.

Tuesday’s results were mixed but not concerning, says Tom Essaye, president and co-founder of The Sevens Research Report.

“The results of those earnings were mixed, but none of the reports provided the type of surprise that would have impacted the broader market and the S&P 500,” Essaye said in a note emailed to The Epoch Times.

Oct. 14 was better, with Morgan Stanley, Progressive, and State Street beating Wall Street expectations on earnings per share.

Blackstone will be the next major investment manager to release its performance from the past three months.

The consensus estimate points to $1.46 in earnings per share and $3.65 billion in revenue, according to Trading Economics.

Read full story here…

Received — 8 October 2026 ⏭ Technocracy News
Received — 7 October 2026 ⏭ Technocracy News
Received — 6 October 2026 ⏭ Technocracy News

AI Is Creating “Doom Loops” All Over the Economy

Technocrats are stealing the world’s data for their own AI training advantage, while killing the golden goose in the process, as this article makes clear. That lack of ethics and morality is seen everywhere in the world of Technocracy. For instance, read the Crucible of Tokens to learn about the biggest heist of private property in world history. They have no conscience about doing anything wrong, and that is the problem. Read up on psychopathy! ⁃ Patrick Wood, Editor.

Recently unredacted legal documents and depositions in the New York Times‘ lawsuit against Microsoft and OpenAI revealed something extraordinary: executives knew full well of the destructive powers of AI when they rushed to launch products featuring the tech in the wake of the explosive success of OpenAI’s ChatGPT.

Internal Microsoft documents quoted by the newspaper’s lawyers warned of an ugly reality now playing out across the entire economy: “doom loops,” in which AI monetizes the work of human creatives while undermining the economic systems that funded their work in the first place.

“Our AI content strategy has started a ‘doom loop’ that will hurt the performance of our models and the entire web at the same time,” one internal Microsoft document reads. “It is highly unusual that an end-product threatens the economic foundations of its essential suppliers, but that is the situation we have created for our LLM business with respect to its ‘content supply chain.’”

The closer you look, the more you see doom loops everywhere. Journalists, for example, used to publish work under the assumption that online platforms would drive readership to them. But now, tools like ChatGPT and Google’s Gemini simply summarize their articles without sending readers to their sites, sparking a crisis that seems poised to wipe out the independent media.

Similar stories are playing out across other industries, from music to filmmaking. Writ large, many are concerned that AI will eventually be strong enough to wipe out huge numbers of jobs, creating a situation in which unemployed consumers won’t have the funds to support the economy.

The irony, as the Microsoft document pointed out, is that the AI industry is a snake eating its own tail: the very products that it’s pushing on users rely on fresh cultural content even as they destroy the economic incentives to create it.

It’s also a dire situation for the open web, which recently celebrated the macabre turning point of being dominated by more AI slop than human-authored content. As companies — most notably Google — are fundamentally changing the way users interact with the web, more and more newspapers, media outlets, and local news stations are seriously struggling to make ends meet.

Worse yet, as the legal documents in the NYT‘s copyright lawsuit showed, insiders admitted that the entire operation is built on theft.

“Millions of people around the world will soon consider large models ‘hoovering up’ all their work to be an astonishing theft of unprecedented proportions,” an internal Microsoft document admitted.

Plenty of other creative industries are also feeling the pressure as AI-related doom loops continue to undermine livelihoods. According to the latest data from the US Bureau of Labor statistics, it’s a bloodbath — with more than 200,000 jobs in “creative” industries having been lost over the last four years alone, roughly the time since OpenAI launched ChatGPT.

AI-generated music is rapidly crowding out human-produced tracks on streaming services, forcing platforms to figure out where to draw the line. Some have implemented hard and fast rules, while others continue to allow users to exploit algorithms with slop.

Hollywood studios are also doubling down on AI while protesting actors struggle to have their voices heard. Many of them fear that their likeness or voice could soon be rendered worthless, representing an existential threat to their livelihoods.

Even the job market is embroiled in its own doom loop. More and more desperate job seekers are turning to AI to brush up their resumes and manipulate AI-powered job portals, while recruiters are drowning in a wall of almost identical applications.

“We’ve got this tragic situation where each side has a problem,” applicant tracking system company Greenhouse CEO Daniel Chait told Wired. “They’re using AI to solve their own problem, but in ways that make the problem worse.”

“And so more AI use begets more AI use, to no one’s benefit,” he added. “The more it’s happening, the worse it gets.”

Similarly, the use of AI in the classroom has skyrocketed, with students using the tech to complete their homework, offloading their cognitive faculties to the machine. Meanwhile, their teachers are using AI to come up with assignments, slowly effacing the human in the loop. Educators who bristle at the idea of using the tech are finding themselves at odds with school administrators.

The effects AI has had on education has been nothing short of a disaster. A recent study by the Brookings Institution’s Center for Universal Education concluded that the risks of using generative AI in education massively outweighed the benefits. The associated dangers have never been more apparent, with educators warning that college students are rapidly losing the ability to read.

Proponents of AI, including OpenAI and Microsoft’s lawyers in the ongoing New York Times lawsuit, argue that AI brings something new to the table. They argue that the output of AI models is sufficiently differentiated from the original content it draws on, falling under “fair use,” a longstanding doctrine in US copyright law.

Whether this argument will hold up in court remains to be seen. Besides, we’ve already seen companies, like Anthropic, settle for billions of dollars after being found to have violated copyright laws.

But considering how the tech is simultaneously undermining the very economic systems that produced the data they need to exist in the first place, that may not be a winning strategy in the long run.

By ripping out the carpet from underneath many parts of the economy — and reaping all the financial gains from it — the AI industry is trapping itself in a doom loop of its own creation, a cynical strategy that won’t benefit anybody but those at the very top who kicked the ladder down behind them.

Read full story here…

Received — 5 October 2026 ⏭ Technocracy News

Elon’s “DeathStar” Starlink Service Will Beat Fiber Everywhere

4 October 2026 at 23:35
Musk owns space. With the launch a new class of 100,000 satellites, he is on the verge of making all fiber-optic networks on planet Earth obsolete while moving Grok AI into space. This will give him the reins over the global communications backbone and a monopoly on information flow, pricing, and access. Does Musk hold the trump card? It sure is sounding like it. ⁃ Patrick Wood, Editor.

Elon’s 250 kW Starlink satellites will bring multi-gigabit satellite internet and 5G satellite cellphone everywhere even high density cities. 250 kilowatts per satellite, 10 terabits per second in both directions, and a path to more than 100 terabits per second.

I detailed the 250 kilowatt per Starlink satellite as beating the Space Station in my article and youtube video from 9 days ago.

WHY build satellites this powerful? Serving everyone and leveraging the huge solar and thermal radiators of the planned AI satellites.

Read full story here...

💾

Elon’s 250 kW Starlink satellites will bring multi-gigabit satellite internet and 5G satellite cellphone everywhere even high density cities. 250 kilowatts p...
Received — 3 October 2026 ⏭ Technocracy News

Naughty Or Nice: UK Officials Are Keeping LIST Of Brits Who Question Government Policy

2 October 2026 at 18:50
My forecast that Free Speech would be dead by year-end 2026 is on target. The battle is lost in the UK, With 34 arrests per day and a whole-of-government policy of tracking dissent, the whole machine has pivoted from tracking terrorist threats to tracking everyday citizens. This is the certain end-state of multiculturalism, started by Peter Sutherland, a member of the Trilateral Commission, when he served as Special Envoy to the UN from 2006-2017.

In the US, companies like Palantir have amassed mountains of data on each US citizen. This data contains demographic and psychographic information, including social media posts, political activism, etc. This data is waiting for the right moment to be weaponized against American citizens. ⁃ Patrick Wood, Editor.

The British state is compiling a list of people who criticise its own counter-terror programme. Ordinary X and Reddit users who questioned Prevent training, flagged a one-sided focus on the “far Right,” or simply criticised it have had their posts logged, stored and “investigated” by a Home Office-linked unit most of them have never heard of.

Officials spent more than a year trying to keep that database secret. They only released it after an appeal to the Information Commissioner’s Office. The public was never told their feeds were being trawled.

The Standards and Compliance Unit, or StaCU, was set up in February 2024 as an “opaque” body to handle complaints about Prevent, the safeguarding scheme sold as a way to stop people becoming terrorists. Documents obtained by The Metro and released by Rights & Security International show StaCU did more than wait for formal complaints.

“Social media posts criticising government policy are being listed, monitored and investigated by officials.”

The government are supposed to serve the public rather than monitor those who disagree with them by Orwellian surveillance and snooping. https://t.co/mH62tNEkYV

— James Melville 🚜 (@JamesMelville) September 30, 2026

Between March 2024 and February 2025 it made 77 “observations” of critical online comments, most of them from X, with others from Reddit and even articles in the Guardian and the Telegraph.

Jacob Smith, Freedom of Expression and Belief Team Leader at Rights & Security International, fought for a year to force the Home Office to publish the material.

“It is shocking that the government has been trawling X and Reddit to find out who has been critiquing Prevent – and then storing that information,” he said. “You should be allowed to criticise government policy without being put on a list.”

He added: “People should be able to express their views on social media without that being catalogued in government databases.”

Many of those logged, he said, had posted without thinking “that the government might be looking at their feed and taking that data. It is in an almost private setting.”

His conclusion was blunt: “It’s a stark reminder that, in the government’s eyes, anything you share online is fair game and could go into a file forever.”

A large share of the logged posts accused Prevent training of treating the “far Right” as the main threat while skating over Islamist extremism.

One X user was recorded after calling the training “complete bullshit.” They wrote: “Anyone’s who done the prevent training knows it only talks about the far Right. Islam extremism isn’t even mentioned.”

Another poster said they “sat through Prevent training at work … and the narrative was very much that the far Right is the biggest threat.” They added: “I kid you not!”

StaCU says it tries to independently verify claims about Prevent. It managed that for only 10 percent of the posts. The unit’s parent body, the Commission for Countering Extremism, sometimes replies in public, inviting the user to submit a formal complaint.

In one case, after a post accused Prevent training of “turning a blind eye to Islam,” the Commission wrote: “Hello, we process and investigate complaints about Prevent and are interested in finding out more about the training referenced above. Please feel free to DM us, tweet us back, or make a formal complaint using this link.”

What that reply does not say is that the user’s post has already been stored in an internal government list.

It was not only anonymous accounts. Open Rights Group had a tweet recorded that said: “How data is managed and stored under the Prevent programme lacks transparency.”

Sarah Lasoye, the group’s Programme Manager for Pre-Crime, told The Metro: “Prevent has expanded far beyond its stated aim of identifying people considered ‘vulnerable to radicalisation’, and has increasingly been used to justify extensive surveillance.”

She added, “It is deeply concerning to see the Prevent duty being used to enable the monitoring of people and organisations simply because they are critical of the programme. That raises serious questions about freedom of expression and the right to dissent and protest.”

“There is also a troubling lack of transparency about what happens to the information gathered through this social media monitoring: how long it is retained, who it is shared with, and ultimately what the Government is using it for,” Lasoye urged.

Rights & Security International says organisations including Hope Not Hate and Maslaha also appear in or around the observations. The Campaign Against Antisemitism had a post logged that did not even name Prevent. It criticised police for failing to take a Hezbollah vigil seriously. A spokesperson said: “We aren’t sure whether to be flattered or concerned that StaCU has picked up a post of ours.”

The group added, “If it’s because we’ve dared to be critical of law enforcement’s lax attitude toward Islamist extremism in this country, we stand by what we said and would reiterate it for as long as it takes to bring about change. But if our post was flagged because StaCU agrees with us that police forces have failed to train their officers on what Islamist terrorism looks like and which organisations are banned, we hope that the message is being received and change is coming.”

That last point tracks a long-running complaint. Sir William Shawcross, in his 2023 review of Prevent, wrote: “The bar for what RICU includes on Islamism looks to be relatively high, whereas the bar for what is included on the extreme Right-wing is comparably low.”

Home Office figures for the year to September 2025 showed 10,293 Prevent referrals. Twenty percent were recorded as extreme right-wing ideology. Eight percent concerned Islamist extremism.

A Home Office spokesperson said: “Prevent does not track people who may have criticised the programme, and nor does it target any particular community.” The department added that keeping the public safe is “our number one priority” and that Prevent has “moved more than 6,000 people away from violent ideologies since 2015.”

The documents say otherwise on the first claim. StaCU collated the posts as “open-source complaints,” stored them internally, and the Home Office refused to release the list until the Information Commissioner forced its hand. Nobody outside the unit can say who else sees the file, how long it is kept, or whether it is shared with police, platforms or other departments.

This is not an isolated compliance desk. It sits inside a growing stack of state machinery built to police speech, “narratives” and online anger.

In June, we laid out how the Home Office’s Research, Information and Communications Unit has been used to manage the mass-migration story — briefing police to cast concerned citizens as “unsympathetic thugs,” shaping family statements after migrant-linked killings, and running a higher bar for Islamism than for the so-called extreme Right.

The same week, ministers moved to give Ofcom faster powers to block “false information” during so-called crisis events — a phrase wide enough to cover any unrest the government would rather you not describe accurately.

London Mayor Sadiq Khan had already demanded a central government social-media “disinformation” unit, complaining of an “outrage economy” while knife crime, robbery and theft climbed on his watch. “If platforms fail to act, the state must have the tools to make them,” he said.

Last week Prime Minister Andy Burnham used his first United Nations speech to announce a National Centre for Information Defence — a new machine to “detect, attribute and disrupt” what ministers call hostile information attacks and to stop a “distorted and untrue narrative about Britain.”

He wrapped it in Russia, bots and “community cohesion.” Nigel Farage called it a Ministry of Truth. “Never ever trust this authoritarian Government,” he said. Defence Secretary Wes Streeting insisted: “We’re not interested in policing domestic political dissent.” Labour’s record is why that promise is already worn thin.

Police have their own version. The National Internet Intelligence Investigations team, stood up after the 2024 Southport riots, has referred more than 100 “suspicious” posts to local forces, including dozens tied to protest-related activity.

Officials describe it as “a dedicated function at a national level for exploiting internet intelligence.” Reform UK leader Nigel Farage called it “the beginning of the state controlling free speech.”

The arrest figures sit underneath all of it. Between 2021 and 2025 at least 62,199 people were arrested in Britain on suspicion of communications offences — section 127 of the Communications Act 2003, the Malicious Communications Act 1988, and the Online Safety Act’s false-communications offence.

That is roughly 34 arrests a day. Big Brother Watch director Silkie Carlo said speech policing is “out of control” and called the result an “Orwellian mess.” Lord Toby Young asked why authorities spend so much time “policing our tweets when they could be policing our streets.”

Prevent is sold as safeguarding. The logged posts are not bomb plots. They are workers calling official training biased, campaigners asking how data is stored, and a Jewish group complaining that police treat a banned terrorist organisation too lightly.

If that is enough to enter a Home Office database, the category of “acceptable” speech is already whatever officials decide it is on a whim.

Britain does not need another unit to hunt “untrue narratives.” It needs a government that can survive being questioned without opening a file on the questioner.

Read full story here…

Received — 1 October 2026 ⏭ Technocracy News

King Of AI Behind Donald Trump: UAEs Sheikh Tahnoon bin Zayed Al Nahyan?

Zayed Al Nahyan is at least an influencer over Trump, if not the puppeteer. This is the “silent partner” in Trump’s World Liberty Financial with a 49% interest, which was hidden from the American public for almost a year. He figured prominently in my book, The Economics of Technocracy, as the Sheik who wants to dominate AI and Fintech globally. His wealth, power, and global influence make Trump look like a wannabe. He is the nexus between Islamic finance and the privatized asset-based tokenized system based on stablecoins. ⁃ Patrick Wood, Editor.

On the night of 12 November 1993, a slim, young Brazilian named Royce Gracie stepped into an octagonal cage in Denver, Colorado, and found himself staring up into the eyes of a giant. Gerard Gordeau, a Dutch kickboxer, stood five inches taller, weighed almost three stone more than him, and had an extra decade of fighting experience under his belt. Now the pair were about to engage in bare-knuckle combat: no judges, no separation of opponents by size, no respite until someone either got knocked out cold or begged for mercy.

The fight lasted barely 100 seconds. In that time, Gracie calmly tripped Gordeau up, jumped on his back, and clutched at his throat, forcing the big man to hammer the floor in submission.

About 80,000 pay-per-view spectators watched Gracie’s legendary victory – the finale of the first ever Ultimate Fighting Championship – live on television. Among them was an unassuming business studies student living in San Diego, California. Mesmerised by what he had seen, the student presented himself at a local martial arts academy that specialised in Brazilian jiu-jitsu – the discipline employed by Gracie to humble Gordeau in the cage – and asked to begin lessons. His classmates found it hard to take the newcomer seriously. “I was there the day he showed up,” said Kid Peligro, who went on to become one of the biggest jiu-jitsu stars in the world. “He was wearing shorts, flip-flops and a tank-top; he sat in the corner and barely spoke to anybody.”

The student, who introduced himself as Ben, was in his mid-20s and hailed from the Middle East, but revealed little else about his background. Peligro assumed that he served as the butler or bodyguard to the rich Englishman who sometimes accompanied him to lessons. “Ben was quiet, and he was dedicated,” Peligro told me. “He would often stay behind after class to sweep the mats.” Ben also took physical pain in his stride, even when it involved fractured bones. “If you trained in jiu-jitsu in the 1990s, you were going to get hurt,” said Bobby Razak, a film producer and martial arts enthusiast. “It wasn’t like it is now: very clean, very professional. Back then, it was raw.”

After more than a year of training, Ben made a startling announcement: he was about to return home to Abu Dhabi, capital of the United Arab Emirates, and would be taking the academy’s founder, Nelson Monteiro, along with him. Other invitations soon followed. Rumours spread that Ben was trying to establish his own martial arts centre on the shores of the Persian Gulf. One day, Peligro and his wife were sent a pair of first-class flight tickets to the UAE, worth $24,000 each. When they stepped off the plane, they were astonished to see pictures of a familiar-looking man – clearly Ben’s father – everywhere they turned, from billboards to banknotes. “Ben’s training centre turned out to be one of his family’s palaces,” Peligro said, still marvelling at the memory. “It’s like a fairytale.”

For those trying to uncover the truth behind His Highness Sheikh Tahnoon bin (or “Ben”) Zayed Al Nahyan – royal fixer, reclusive spymaster, financial behemoth – fairytales are nearly all we have to go on. Tahnoon, who really did enrol in Nelson Monteiro’s class under a pseudonym and eventually went on to become a jiu-jitsu black belt in his own right, now sits atop an economic fortune larger than the GDP of several major countries, and is closely entwined both personally and financially with the Trump White House. The global holdings overseen by Tahnoon span windfarms in Scotland and spaceships in Texas, Rihanna’s lingerie line and Richard Caring’s Ivy restaurants. From hospitals to housing blocks, solar farms to ski resorts, there is almost no place on Earth untouched by his financial firepower.

But despite his ever-growing influence on the world stage, Tahnoon prefers to remain in the shadows. He has never given a substantive interview to the press. Those who have worked closely with him are almost always made to sign non-disclosure agreements. Very few people outside the UAE even know his name. “He’s quiet,” says one former US national security official, “and he’s always everywhere.” Sheikh Tahnoon did not respond to an interview request for this article. The UAE foreign ministry did not offer any other senior government figure for interview, and declined to provide any comment.

Tahnoon’s power is soon going to extend over more human lives than ever before. As one of the most important AI financiers on the planet, he is rapidly embedding himself within the infrastructure that will underpin our digital future: its information pipelines and its microchip foundries, its datacentres and the enormous energy plants needed to power and cool them. Some see him as a technological prophet and saviour. Others compare him to a Bond villain. So who is the real “Ben”, and what are his plans for the rest of us?

Anyone trying to find Sheikh Tahnoon must first navigate concentric circles of gatekeepers. These include his closest business associates, as well as a cast of more colourful characters: the former Lebanese prime minister Saad Hariri, for example, or Melissa Moncada, one-time winner of the Miss Globe Colombia beauty pageant, and now a fitness entrepreneur. Even if you get this far, it may not be enough to be granted an audience, and supplicants go to extraordinary lengths to stand out from the crowd.

These scattered landmarks form the backstage of the modern Emirates. Together, they also tell the story of the six brothers who have come to dominate the country – a story that began more than half a century ago, with their father, Sheikh Zayed bin Sultan Al Nahyan, the UAE’s first leader following independence from Britain in 1971. During his 33-year reign, a loose network of Bedouin tribespeople living among mud-brick forts and palm-frond houses was transformed into an oil-rich regional superpower, home to some of the tallest buildings on Earth.

Zayed married seven times and is believed to have fathered at least 30 children. But it was the offspring of his third and favourite wife, Fatima Al Ketbi, who – under her tutelage – emerged as their nation’s powerbrokers. Tahnoon was one of them. Today, he and his siblings – known as the bani Fatima, or “sons of Fatima” – sit at the apex of arguably the wealthiest dynasty in the world.

Tahnoon and his brothers grew up steeped in a worldview that placed their family at the centre of the universe. Any form of dissent was seen as an existential threat, not just to their own power but also to the small and fragile state they were responsible for protecting. “The family is the first school of values, the authentic guardian of identity, and the truest source of trust and belonging,” their mother, Fatima, once said, reflecting both the paternalistic ethics that prevail among the Emirati elite, and the Al Nahyans’ visceral hostility towards alternative ideologies such as socialism or political Islam. By the time Tahnoon returned from the US in 1995, a state-sponsored crackdown on Islamist civil society groups was under way, alongside reprisals against smaller emirates that resisted it.

During this period, the UAE’s leadership became increasingly conscious that money alone could not buy them international clout. Despite oil finds in the Gulf, it was the big Arab republics such as Egypt and Syria that still dominated the Middle East, both diplomatically and culturally. Military strength and strategic influence were needed to propel the Emirates into the big time. In the final years of his reign, Sheikh Zayed ensured that the bani Fatima were favourably positioned to drive this project forward.

As a younger brother, Tahnoon knew that the Abu Dhabi throne – and consequently the presidency of the UAE, which always accompanies it – was not destined for him. Instead, he was given the task of running his father’s private department, which oversaw Sheikh Zayed’s extensive personal assets, including palaces, farms and overseas landholdings such as London’s Berkeley Square. The role demanded a blend of charm and coercion, and the deployment of hard cash. By the late 1990s Tahnoon was making the latter his speciality, founding a company called the Royal Group that became the primary investment vehicle for his family’s many business interests. The group’s assets are currently worth $300bn.

Following Sheikh Zayed’s death in 2004, Tahnoon’s reputation as the UAE’s top money man grew. As comfortable in the boardrooms of Wall Street as he was in the Emirati majlis (the traditional court of the tribal ruler), Tahnoon melded the rhetoric of Bedouin values – duty, loyalty and humility – with a new vocabulary of technocratic managerialism. Today, the country’s largest sovereign wealth vehicle is run by Tahnoon; so is First Abu Dhabi Bank, its largest lender, and Alterra, the biggest climate investment fund in the world. In the UK alone, Tahnoon oversees healthcare providers and property giants, and controls investments in public utilities such as Thames Water. The economic empire he oversees is estimated to be worth about $1.5tr, as large as the entire economic output of Indonesia or Turkey.

But the line between Tahnoon’s personal fortune, the wider wealth of the Abu Dhabi royal family, and collective assets held – at least in theory – by the citizenry of the UAE, is blurry. According to the bani Fatima, that division is moot, as the wellbeing of the Al Nahyans is indistinguishable from the wellbeing of the nation at large. Not everyone agrees. “A lot of merchants in the UAE complain constantly – very quietly, very privately – about how Tahnoon crowds them out of the market,” Mira Al Hussein, an Emirati sociologist based at the University of Edinburgh, told me. In healthcare, for example, sovereign wealth funds and Tahnoon’s own corporate vehicles dominate the local industry; the private sector “cannot compete”, one senior executive recently told the Financial Times. “The general perception is that the country is a ship,” said Al Hussein, “and if that ship ever begins to sink, then the ruling family will jump off and take all of the wealth funds and investments and portfolios with them.”


Tahnoon is obsessed with winning. In the late 1990s, the sheikh became so determined to beat his brothers at the newly released video game Age of Empires – in which players attempt to transform pre-modern societies into monumental civilisations by hoarding resources and overpowering opponents – that he hired one of the game’s software developers to coach him. By the early 2000s, his attention had switched to chess. He personally financed the development of a refrigerator-sized supercomputer called Hydra, which soon became the most powerful chess player in the world, beating several grandmasters along the way. Hydra could only ever be used by a single person at a time. For hours at a stretch, day and night, that person was Tahnoon. In 2005, Hydra’s chief architect claimed in an online chess forum that Tahnoon was the greatest computer-chess freak alive.

Tahnoon has brought all these skills – fierce competitiveness, the careful accumulation of positional advantage, and an abiding fascination with melding human ingenuity and computing intelligence – to the financial realm. They have also proved vital in his other main role: as the man in charge of national security in the Emirates, and his country’s spymaster-in-chief.

Many of the individuals targeted by Tahnoon’s security apparatus can’t share their testimony. The UAE has a long history of enforced disappearances, secret detention sites and vicious reprisals against the family members of exiled dissidents – all extensively documented by international human rights organisations. But some survive to tell their tale.

In April 2018, Matthew Hedges, a young British academic, travelled to the UAE to conduct research for his PhD exploring the country’s national security strategy, a subject that touches directly on Tahnoon’s domain. By the end of his two-week visit, Hedges had started to feel uneasy. “I knew I was being followed,” he told me recently.

Entering Dubai airport on his way back to the UK, Hedges got no further than passport control. There he says he found a large group of masked men waiting for him, all dressed in black and carrying guns. He alleges that he was blindfolded, handcuffed and thrown in the back of an SUV. He was accused of working as a British spy and over the months of solitary confinement that followed, he says he was subjected to interrogations lasting up to 15 hours a day and made to swallow a cocktail of medications against his will. His captors allegedly threatened him with rendition to an overseas military base, and warned that they could harm his relatives. Following seven months of maltreatment, Hedges attempted to take his own life. “I knew their threats to be meaningful, because I knew the country,” he told me.

In the room where he was questioned by officials, Hedges says he saw bloodstains on the carpet and bloody footprints on the walls. “They were above my head height, and facing down,” he told me. “I’m over 6ft tall, so that means someone had been lifted up there.” When I asked Hedges who was ultimately responsible for his ordeal, he ran carefully up the chain of command within the Emirati security services, reciting names and job titles, before sitting quietly for a moment and taking a breath. “Then you get above the clouds,” he replied, eventually. “Then there is Tahnoon.” (In a government statement issued in 2023, the UAE insisted that Hedges “received entirely proper care … He was never subjected to, or threatened with, either torture or cruel, inhuman or degrading treatment of any sort.”)

Hedges ultimately confessed to a series of crimes, but insists that he did so under duress. In November 2018, following a hearing that lasted only five minutes, at which there was no defence lawyer present, he was sentenced to life in jail by an Emirati court. Following a diplomatic row with the British government, he was later granted a royal pardon. After his release, the UAE government mounted a smear campaign against him that included the release of a 19-page dossier of personal information including a psychiatric report. Compared with many others who pass through the country’s interrogation cells, he got off lightly. “Hedges’ experience is unfortunately common in the UAE, and abuses against Emirati civil society are even more repressive,” said Joey Shea, a senior Gulf researcher for Human Rights Watch.

Supporters of the Emirati regime bridle at this sort of criticism. They insist that it is grounded in western resentment at the UAE’s rise, and overlooks the unique security threats the nation faces from hostile neighbours and ideological insurgents. “The UAE has become a success story, and that’s led to you guys thinking, ‘Wait, these were the little ducklings that we used to run and now look at what they’re doing,’” said Abdulkhaleq Abdulla, an Emirati professor of political science who is well connected to ruling officials in the UAE. “I resent the claim that the UAE must ‘level up’ to international standards [on human rights]. You have your standard, we have ours.”

The Al Nahyan dynasty has good reason to feel insecure. Located on the tip of a landmass roiled by conflict to the north and south, and staring out across the Gulf at its far larger rival Iran, the UAE’s petro-riches make it an obvious target. Tahnoon and his brothers also know that their brand of autocratic monarchy is despised by political Islamists and pro-democracy Arab revolutionaries alike. Some 20 years ago, Christopher Davidson, an expert in Gulf politics at Durham University, spoke to one of the bani Fatima about the Emirates’ rising prestige. “He said to me, ‘You need to understand the mindset of this region. Others will try to take that away from us, or destroy us’,” Davidson recalled. “‘Whether it’s Iran, Yemen, the Muslim Brotherhood, Al-Qaeda … they’ll all be coming, one after another.’”

For the UAE’s leaders, the 9/11 attacks – perpetrated by a group of extremists that included two Emiratis – were a realisation of their darkest fears about violent enemies lurking within, and helped fuel an obsession with Islamism that persists to this day. (The UAE has been linked to covert funding and political support for far-right political parties in Europe who are seen to share Abu Dhabi’s perspective on this issue.) The US war on terror was also a moment of opportunity, enabling the Al Nahyans to present themselves to Washington as an island of stability within an otherwise forbidding region.

Throughout the 2000s, the Emirati regime ramped up its intelligence activities and increasingly focused on espionage technologies. The aim was to mitigate the country’s most obvious vulnerability: its tiny native population of 1.4 million people, which comprises only 11% of the UAE’s total residents, and makes it hard for the country to raise a significant army or to function in the absence of foreigners. By the end of the decade, it had struck a landmark security deal with the former US counterterrorism czar Richard Clarke, who was instrumental in lobbying Congress for more intrusive surveillance programmes both at home and abroad. Clarke moved out to Abu Dhabi and set up a secretive UAE spying unit, which was given the codename Project Raven.

Initially billed as a vital cog in the “war on terror”, Project Raven morphed into a vast and pervasive digital wiretap operation targeting not only Emiratis accused of terrorism, but many others merely suspected of being unfriendly to the UAE’s dictatorship. Although technically an Emirati initiative, much of the team running Project Raven were former US National Security Agency spies, Washington contractors and senior White House veterans. “The incentive was to help in the fight against terrorism,” Clarke later insisted. “The UAE is a very good counterterrorism partner.”

By 2011, when anti-government uprisings erupted across the Arab world, the focus of Project Raven’s energies had become clear. After a group of prominent Emirati judges, lawyers and journalists put their names to a petition calling for mild democratic reforms, the state’s response was uncompromising; over the next two years the majority of them were arrested, and dozens of signatories remain in jail today. More paranoid than ever, and harbouring fears that Emirati soldiers might refuse to turn their guns on fellow Muslims in the event of widespread street protests, the bani Fatima mobilised an 800-strong paramilitary force of Colombian and South African mercenaries to tighten their grip on power, and even greater resources were funnelled into digital hacking operations. To the Al Nahyans, mass politics was a threat: in their eyes, grateful subjects could only ever petition their leaders for concessions, never make demands of them as citizens. “The UAE’s internal security structure increasingly prioritised regime protection above all else,” said Mira Al Hussein.

It was against this backdrop that Tahnoon – alongside his role as the UAE’s money man – began rising up the ranks of the Emirati security establishment. In 2014, he became the country’s deputy national security adviser; two years later, he assumed the top job. As security chief, his new brief included some of the UAE’s most sensitive diplomatic missions, among them the breaking of a decades-long taboo in the Arab world by normalising relations with Israel. As the Middle East’s other leader in the field of cyber-espionage, Israeli intelligence had much to offer the Emiratis: not least access to Pegasus, the most advanced spyware ever built. With Pegasus, operatives were able to infiltrate almost any Apple or Android device without the user’s knowledge, and transform it into a permanent, pocket-sized spy.

Under Tahnoon’s stewardship, Project Raven – control of which had now passed into the hands of a private, state-linked Emirati company named DarkMatter – scaled up its hacking and surveillance activities. Local dissidents and suspicious foreigners on Emirati soil were targeted. But so, too, were human rights activists, journalists and politicians all over the world. Fiona Shackleton, a high-profile British divorce lawyer representing an estranged wife of the ruler of Dubai – whose other clients have included Prince Charles and Paul McCartney – was one such victim. Then UK prime minister Boris Johnson is believed to have been another. It seemed that no matter who you were, if you were deemed to threaten the interests of the Al Nahyan family and their friends, even tangentially, then you were a potential target.

In early 2019, a Reuters investigation – building on work by the Canadian research group The Citizen Lab – exposed much of Project Raven’s operations, and multiple legal proceedings were launched in the US against the initiative’s American contractors. Over the next year, Tahnoon officially shut down DarkMatter and broke apart its various entities. By that time, it appears he was nursing more ambitious plans.

A powerful new frontier in computing power – artificial intelligence – was just beginning to open up. To be genuinely transformative, this technology was going to require fantastical amounts of cash and data. By fusing together elements of his business and security empires, Tahnoon would be in a unique position to supply both. A company like no other was about to be born.

In Douglas Adams’s The Hitchhiker’s Guide to the Galaxy, a superintelligent computer programme named Deep Thought is asked “the great Question of Life, the Universe and Everything’”. After seven and a half million years, it finally delivers its famous answer: 42. Deep Thought goes on to say that, in order to explain its answer, it needs to design another giant computer programme. This one will be called “Earth”, and one of its core operational components will be human beings.

In this vision, the number 42 represents more than the meaning of life. It speaks to a world in which every facet of reality can ultimately be reduced to lines of code, and where ordinary people unwittingly live out their days in the service of computers rather than the other way around. In the late 2010s, when Tahnoon refashioned the ruins of DarkMatter into a new AI-focused company, he decided to honour Adams by calling it Group 42, or G42 for short.

G42 is the culmination of Tahnoon’s dual identity as a financier and spymaster, and the scale of its aspirations is staggering. Like many other tech giants, it aims to reconceptualise AI as a foundational form of infrastructure like water or electricity – a concept known as the “intelligence grid”. Unlike its Silicon Valley peers, however, G42’s intelligence grid has grown out of an actual nation state’s intelligence apparatus. The company’s marketing heavily emphasises its security capabilities, promising “real-time threat defence” for governments and businesses, and predictive tools that encompass “suspect profiling”, “relationship mapping” and “behavioural analysis”.

The tech industry is not short on exaggeration and bombast. Compared with many of his Silicon Valley peers, however, Tahnoon – a true AI believer who claims to have designed the entirety of one of his homes via ChatGPT – is in a unique position to deliver. G42 built on the espionage talents behind Project Raven and DarkMatter: Peng Xiao, who was the head of a key DarkMatter unit, is now G42’s CEO; Karim Sabbagh, DarkMatter’s former boss, currently runs one of G42’s main subsidiaries. One of G42’s first public products was a popular messaging app called ToTok, which was later revealed by the New York Times to be a piece of spyware, designed to “track every conversation, movement, relationship, appointment, sound and image of those who install it on their phones”.

In a statement to the Guardian, G42 denied that the company’s origins lay in cyber-espionage operations, or that it conducts or enables unlawful surveillance. “G42 was established as an AI and technology company focused on cloud and AI infrastructure,” the firm said. “Our businesses operate under applicable laws, contractual data protections, and rigorous security and compliance frameworks.”

If US tech giants are wary of G42’s links to cyberhacking, they are doing a good job of hiding it. Microsoft has invested $1.5bn in the company and its vice-chair, Brad Smith, now has a seat on G42’s board of directors. G42 and its specialist AI investment fund – also chaired by Tahnoon – are either direct investors in or major partners of Anthropic, xAI and OpenAI. Sam Altman, the latter’s CEO, calls Tahnoon “a dear personal friend”. Microsoft and OpenAI were approached for comment on this article, but neither company provided a response. In its statement to the Guardian, G42 noted that its partnership with Microsoft “is governed by a binding agreement developed in consultation with the US and UAE governments, covering cybersecurity, data protection, responsible AI, export controls and business integrity”.

Tahnoon’s appeal to Silicon Valley is obvious. After all, if you’re going to go into business with a cash-laden Gulf autocrat, then picking one who presides over a heavily digitised state-security empire – and who boasts a proven track record of hoovering up humongous amounts of personal data, the raw ingredient upon which your technology depends – feels like a logical choice. Mark Zuckerberg, Sam Altman and Elon Musk may all be hugely influential AI leaders, but none of them simultaneously runs the CIA.

Another powerful figure to have clutched Tahnoon and G42 close to his bosom is Donald Trump, who has hosted lavish state banquets for the sheikh at the White House and posed for photos with him in the Oval Office. The American security services have long harboured fears about Emirati influence on US politics: in 2022 the National Intelligence Council produced a confidential report for Washington policymakers warning of Abu Dhabi’s “illegal and legal attempts to steer US foreign policy”. Joe Biden’s administration restricted the export of AI chips to the UAE because of concerns over the country’s close ties to China. But last year Trump reversed that decision, authorising the transfer of hundreds of thousands of Silicon Valley’s most advanced chips to the Al Nahyan regime. Many of them will go straight to G42.

The deal was announced only two weeks after one of Tahnoon’s investment vehicles acquired $2bn-worth of cryptocurrency issued by World Liberty, a company owned by the Trump family, who earn interest on all its reserves. Tahnoon is also a direct investor in World Liberty, having purchased 49% of the firm less than a month before Trump’s second inauguration. Both the White House and World Liberty have denied any connection between Tahnoon’s investments and the chip transfer. A spokesperson for G42 told the New York Times that the company was “grounded in integrity”.

“Corruption, plain and simple,” is how the Massachusetts senator Elizabeth Warren describes the murky entanglement of Tahnoon’s money and the US White House. But the rapid ascent of Tahnoon and G42 within the AI industry reflects a sincerely held conviction on the part of both the Trump administration and the bani Fatima that AI is set to replace oil as the elemental global commodity – and ultimate determinant of global power – and that now is the time to master it.

The result is that G42, under Tahnoon’s control, has successfully positioned itself at the heart of global AI infrastructure for decades to come. The “AI campus” currently being constructed on the UAE’s coastline is projected to have a 5GW power capacity, up to 250 times the size of the largest hyperscale datacentres today. G42 is helping to finance a $500bn rollout of new datacentres within the US, and has now acquired several of the construction companies involved. Last year, Microsoft and G42 officially co-launched the Responsible AI Future Foundation, designed to shape “ethical development and deployment of AI systems”. It is headquartered – of course – in Abu Dhabi.

Does it matter if significant chunks of our AI infrastructure sit in the hands of a man with the entire arsenal of a police state at his disposal? One person who might be able to answer that is Ahmed Mansoor, a mild-mannered Emirati electrical engineer and aspiring poet. In the early 2010s, Mansoor was among that group of public intellectuals who signed a petition calling for the creation of an elected parliament in the UAE; he also ran a website on which content critical of the Al Nahyan regime was sometimes posted. In April 2011, he was arrested on multiple charges, including the defamation of the bani Fatima, but initially released thanks to a presidential pardon.

In the years that followed, though, Mansoor was relentlessly targeted by the Emirati security services, who spent millions of dollars attempting to gain access to his electronic devices, and those of his wife. Eventually, Mansoor succumbed to a malicious email attachment, which sent all of his private data to a computer server registered to PO box 5151 in Abu Dhabi: the postal address of Sheikh Tahnoon’s personal office. He later found out that his baby monitor had also been hacked, turning it into a camera and microphone that captured every moment of his family life. He was arrested again in 2017, subjected to a secret trial, and convicted of insulting the “status and prestige of the UAE and its symbols, including its leaders”. Sentenced to 10 years in jail, he has remained in solitary confinement in al-Sadr, the maximum-security jail located about 10 miles south-west of G42’s AI mega-campus, ever since.

We don’t know if Tahnoon will use his commanding position at the heart of the AI industry to directly hack, surveil and intimidate critics, as his security services did with Mansoor. But the character of those controlling this new technology matters. Today, we don’t just use major tech platforms to store files and send messages: more and more people use AI to be their therapist, friend or even lover. Tech executives assure us that our data is safe from prying eyes but ordinary citizens have very few ways of verifying this. In its pomp, the early smartphone manufacturer BlackBerry insisted that “privacy and security form the crux of everything we do”. It failed to mention that BlackBerry was secretly working with the cybersecurity unit of a foreign government at the time, and helping it to carry out digital surveillance of mobile users. That cybersecurity unit was the UAE’s DarkMatter group, the precursor to G42.

The potential risks of untrustworthy actors controlling AI may be more insidious than direct hacking and surveillance. A group of researchers recently revealed that AI tools – including products from Meta, Google and SpaceX – are regularly altering the meaning of texts on sensitive political topics, in ways that could reshape public opinion. “I’m very worried about Tahnoon’s perception of the world,” says Al Hussein. “He’s the video-game prince who wants to press a button and eliminate all threats.”

For a glimpse of what a world reshaped in Tahnoon’s image might look like, consider the UAE. Here, political parties are nonexistent, trade unions and labour strikes prohibited, and protests and lèse-majesté met with rapid punishment. Migrant workers, who make up the vast majority of the country’s residents, are kept in a state of legalised insecurity and suffer from thousands of officially unexplained deaths each year, which human rights organisations attribute to heat stress, construction accidents and other poor working conditions. Businesses, by contrast, can pick and choose the regulations that govern them from a patchwork quilt of 40-odd “freezones”, all competing to lure investors with bespoke tax regimes and legal frameworks.

Beyond the UAE’s borders, Abu Dhabi’s dominion encompasses a vast network of ports, security partners, client regimes and financial flows spanning the Red Sea and huge swathes of north, eastern and central Africa. Often presented as a neutral entrepot, torn between rival global power blocs to the east and the west, the reality is that the UAE has always served faithfully as capital’s emirate: shielding the wealth of international elites, opening up inhospitable markets, paving the way for western investors, and – as Colin Powers, a political economist specialising in the Middle East, puts it – “energetically extending the circuitry of global wealth accumulation”.

Under the bani Fatima, this function has included the brutal suppression of any democratic experiment within the Emirati field of influence that might threaten the material interests of the Al Nahyans or their allies. In Egypt, the UAE has helped sustain the military dictatorship of Abdel Fattah al-Sisi, which has turned its guns upon thousands of protesters. In Yemen, Libya and Ethiopia, it has armed paramilitaries and fuelled shocking levels of state violence against civilians. In Sudan, it is accused of complicity in genocide. In response, Abu Dhabi has deployed AI tools as part of a sophisticated disinformation campaign designed to discredit the United Nations. Several dissidents wanted by other dictatorships – from Iran to Saudi Arabia and Rwanda – have also been kidnapped while transiting through Dubai airport, before being “delivered” by the Emiratis to their home country’s security forces. Not all of Abu Dhabi’s foreign adventures have been successful; many have led to military quagmires and reputational damage on the world stage. But for those on the receiving end of Emirati intervention, the consequences have been dire.

The imperial ambitions of Tahnoon and his brothers do not seem to rest on direct occupation of land, but rather on controlling key “nodes” outside their borders: the sites at which the movement of money, arms, trade, technology and ideas intersect. AI, the functioning of which is distributed across territories around the world, is the ideal asset for a nodal empire. Tahnoon’s goal of making the UAE an AI superpower is framed domestically as part of his family’s historic mission to improve the lot of the Emirati people. But the UAE’s accumulation of power in recent decades has been delivered on the bani Fatima’s own terms. Central to the Al Nahyan family’s worldview is the notion that, amid a planet riven by the chaos and humiliation of democracy, their nation is a demonstration that things can and should work differently. According to Tahnoon’s brother, President Mohamed bin Zayed Al Nahyan, “The UAE is a tangible model for the world.”

With his array of spy tools, superyachts and supercomputers, his preference for remaining far from public view, his habit of always wearing dark glasses in public (even indoors), and – of course – his obsession with personal longevity (he aims to live to 150), it would be easy to characterise Tahnoon as a cartoon villain. But in truth, this is not a story about a malevolent prince plotting vile deeds from his palace in the Persian Gulf. It is a far more prosaic tale about how power and money function in the world today.

In Washington, just as in Abu Dhabi, politics is increasingly governed by personal relationships and transactional gains rather than popular demands. In the UK, as in the UAE, firms such as Palantir that offer AI-enabled surveillance and data crunching are being integrated with the basic functions of the state. All over the planet, business-friendly strongmen are the order of the day. “For the last 20 years, democracy has been in decline,” said Ronald Deibert, the founder and director of The Citizen Lab. “That happens to coincide with an enormous expansion of digital surveillance capabilities for both commercial and national security purposes. The epicentre of a lot of this is the UAE. And where do all roads lead in the UAE? They lead to Sheikh Tahnoon.”

On 27 February, Tahnoon chaired a board meeting at G42’s AI investment fund. That night, the sky above Abu Dhabi’s seafront lit up with the flashes of US and Israeli missiles raining down across the Persian Gulf, marking the start of the Iran war. One of Tehran’s first acts of retaliation was to directly bomb two AI datacentres in the UAE. The Iranian government soon published an official list of what they considered “legitimate” military targets which included the Emirati offices of US technology companies such as Amazon, Google and Microsoft. Western journalists speculated that this could spell the end of Tahnoon’s AI dreams. “Why did we ever think datacentres in the Gulf were a good idea?” ran one headline in the Financial Times.

Over the following months, Tahnoon shuttled around the Middle East, conducting covert diplomacy. According to Reuters, senior officials from Iran’s revolutionary guards were hosted by Tahnoon in his Abu Dhabi guesthouse. By mid-June, he had reportedly struck an agreement that provided billions of dollars for the Iranian government, and brought Iran’s strikes on the UAE to a halt. A few weeks later, on the same day that Trump announced that the US’s ceasefire with Iran was “over”, the US commerce department quietly confirmed that it was removing the last remaining export controls restricting the sale of advanced AI chips to Abu Dhabi.

As life in Abu Dhabi gradually returned to something resembling normality, Tahnoon’s family members began reappearing in public. A company founded by his daughter, Sheikha Fatima bint Tahnoon, signed a memorandum of understanding with G42 last month, suggesting that she is being primed to follow in his footsteps and eventually assume a commanding role within the UAE’s tech-security empire. Tahnoon’s son, Sheikh Zayed bin Tahnoon, was the guest of honour at several major jiu-jitsu events this summer. Tahnoon himself, meanwhile, stayed far from the public eye. Judging when to retreat, and knowing when to strike: “Ben” had outmanoeuvred his doubters once again.

Read full story here…

Trump to name AI czar within days: ‘We will have somebody extraordinary’

Every candidate on the short list is mentioned prominently in The Final Betrayal: How Technocracy Destroyed America. What is Trump thinking about establishing the AI Force?  Will this fall under the Department of Defense to be weaponized against Americans? Technocracy is turning ugly and has set its sights on capturing the whole enchilada: the Government, the financial system, the rationing of resources, and control over all citizens. Just like 1930s Technocrats, look for calls to dismiss Congress altogether and appoint Technocrats from top to bottom to run everything in society. Yes, that’s the historical record. Are you willing to risk a repeat? ⁃ Patrick Wood, Editor.

President Trump said Tuesday that he will name his artificial intelligence czar within the next few days.

Speaking at the White House alongside artificial intelligence executives, Mr. Trump said he expected to announce his pick within the next “three-to-four days.”

Mr. Trump said he wanted to seek opinions from AI companies, House Speaker Mike Johnson, Louisiana Republican, and Senate Majority Leader John Thune, South Dakota Republican, before selecting a candidate.

“We will have somebody extraordinary,” he told reporters.

Candidates who have been floated for the position include Sriram Krishan, the senior White House policy advisor on AI; Sean Cairncross, the administration’s national cyber director, and Michael Kratsios, director of the White House Office of Science and Technology Policy.

Treasury Secretary Scott Bessent was said to be among the potential candidates, but Mr. Trump said last week that he eliminated him from consideration because he wanted him to stay at the Treasury.

Industry insiders and some lawmakers have called for limits on the technology because of concerns that machines could go rogue or fall into the wrong hands and cause widespread harm. Mr. Trump dismissed such claims, saying the guardrails need to be relaxed to keep America competitive.

Earlier Tuesday, he hosted a lunch with artificial intelligence executives. Mr. Trump now refers to the technology as “super intelligence.”

Mr. Trump said earlier this month that whoever gets the position will likely lead a new “AI Force.” The president said it would be similar to the Space Force, which he established during his first term, but he did not disclose whether it would be a new branch of the military or how it would work.

David Sacks, a venture capitalist, previously served as Mr. Trump’s AI and crypto czar, but left the White House earlier this year once his term as a special government employee expired.

Read full story here…

❌