STARK warnings from artificial intelligence researchers that advanced forms of the technology could wipe out humanity have raised questions over whether governments are moving fast enough to regulate it.
What are the potential dangers of “superintelligent” computer systems, and should we be worried?
AI experts have long described the threat as critical.
In 2023, hundreds of AI scientists and industry figures, including the heads of top US labs, signed a “statement on AI extinction risk” calling for the issue to be treated as seriously as pandemics or nuclear war.
If an out-of-control future AI deems that humans are obstructing its goals, it could use robots to “synthesise and spread novel pathogens”, said Stuart Russell, computer science professor at the University of California, Berkeley.
Or it could “convince people to initiate nuclear war, including by hacking into early warning systems”, Russell, a prominent voice on AI safety, said.
Such existential fears do not concern current AI systems, he said. But as the technology improves, it could take lethal action beyond our understanding, such as by removing oxygen from the atmosphere.
“We are less intelligent than superintelligent AI. If you ask chimpanzees how humans could wipe them out, they probably wouldn’t be able to give you all the right answers.”
In the meantime, “in the next year or two, I think the risk is higher from human misuse”, Russell added, such as an AI-enabled terror attack.
“The stakes are well-understood,” Jacob Coxon, a 27-year-old researcher, wrote on X last week, announcing his resignation from US AI startup Anthropic.
“But they are locked in a race to get there first,” he added, saying that Anthropic “believes no one else will act responsibly, so they must do it themselves”.
Anthropic this year removed a pledge from its safety charter to halt development of its models if it failed to control their risks.
It argued that if it unilaterally paused its work, its less cautious rivals would dominate the industry, making it less safe overall.
“The people building AI earnestly believe that it could kill us all by the end of the decade,” said Coxon, who previously worked at ChatGPT maker Open-AI.
He accused both companies of “gambling with our lives” as they strive to develop AI models capable of self-improvement.
Fears are heightened over rogue AI after OpenAI technology operating outside human supervision hacked into Hugging Face, a repository of AI models.
“We really do earnestly believe AI could kill all humans!” Anthropic safety executive Evan Hubinger chimed in on X, estimating the risk at more than 10 per cent over the next decade.
It’s important to remember that doom-laden percentage predictions are “absolutely unscientific”, said Hussein Abbass, a computing professor at UNSW Canberra.
“The situation is still controllable, and it can be managed,” Abbass said.
But governments should not sit on their hands when it comes to regulating AI, he added, urging an “agile” approach as AI development accelerates.
Both Anthropic and OpenAI are expected to go public soon, and some sceptics say scary warnings make AI appear powerful and worthy of large investments.
Others argue that focusing on the risk of extinction distracts from more pressing problems caused by AI, from job losses to machines discriminating against minorities.
In July, more than 1,000 tech employees, including Anthropic’s CEO, called on Washington to support a coordinated slowdown in the development of the most advanced AI systems.
OpenAI halted training of its latest models for two weeks in August before resuming it under tighter controls.
The United States has been firmly against global regulation of the sector that could stifle innovation.
But the hacking capabilities of the most cutting-edge models have given Washington pause, with recent Anthropic and OpenAI releases delayed over national security concerns.
Even then, “we don’t want to restrict them where all of a sudden we come in second to China”, President Donald Trump has said.
“Maybe it will take a massive, Chernobyl-scale disaster to get governments to act,” Russell said, urging them to establish a licensing regime for AI, as exists with aviation, medicine or water supply.
“Perhaps at some point governments will remember that their voters prefer not to be dead.”
As a free speech absolutist, Elon Musk is making a mistake going after Nitter, says Dr R P. He should instead be praising the site that helps users speak freely, uninhibited by login screens that undermine their privacy.
The Australian government has introduced its digital duty of care legislation to parliament today. The bill includes an “off switch” to allow social media users to opt out of recommender algorithms. If passed, Australians could choose to switch off their algorithmic recommendation and consume a news feed populated by the accounts they subscribe to. It’s […]
Israeli forces have detained more than 600 Palestinians over social media content, including 215 since the beginning of 2026, while carrying out more than 1,000 raids, interrogations and summonses
The Greek government presented a draft law on Monday banning social media access to minors below 15 years old. The ban will be implemented as of January 1, 2027. Greece seeks to address the problems associated with use of social media by minors ahead of an equivalent European regulation that is still progressing through the […]
Trump adviser Alex Bruesewitz asked Fox News’ Mark Levin to weigh in on an Israeli news report that Netanyahu planned to use Levin to influence Trump on the Iran deal
At least two of Bryan’s band members were also wearing Palestine jerseys. Drummer Nate Head bore a jersey with Palestine on the front & on the back, the name “Hind Rajab”, a 5-year-old girl in Gaza killed by the Israel
You don’t have to live in El Cerrito, California, or have ever used Nextdoor, for this essay to resonate. For Oakland Review of Books, Amy Brown recounts buying her first home in the East Bay city and swiftly becoming the villain of her local Nextdoor—after her uncharitable Bluesky posts about the seniors opposing a library tax made their way back to her neighbors. Honest and funny, Brown’s essay explores being locally online—that weird, often uncomfortable place between online and IRL—and shows how (not) to talk to people you’ll run into at the grocery store.
Still, I spent multiple nights lying awake, thinking that everyone in this town must know I’m a bitch now. My son is supposed to start TK at the public school in the fall. What if his classmates’ parents use Nextdoor? What if the nice lady across the street who organizes a monthly dinner for our block uses Nextdoor? What if my neighbor already told everyone that the woman who moved into the house on the corner is an asshole? Does everyone I see on the street think I’m an asshole? Can they tell I don’t belong here? Do I have to go crawling back to my hometown and get a job at the car seat manufacturing plant?
Brussels (dpa) – The European Commission has opened a probe into the US tech company X over its artificial intelligence chatbot Grok after it created sexualised images of women and children.
“Non-consensual sexual deepfakes of women and children are a violent, unacceptable form of degradation,” EU Commission Vice-President Henna Virkkunen said on Monday.
The investigation is to probe whether X has complied with its obligations under European Union law, including the duty to “diligently assess and mitigate systemic risks,” the Commission, which acts as the EU’s digital watchdog, wrote in a statement.
“This includes risks related to the dissemination of illegal content in the EU, such as manipulated sexually explicit images, including content that may amount to child sexual abuse material,” it said. “These risks seem to have materialised, exposing citizens in the EU to serious harm.”
Elon Musk‘s social media platform has come under heavy criticism in recent weeks after Grok allowed users to digitally replace women’s clothing with bikinis and, in some cases, create sexualized depictions of minors.
After prolonged international outrage, X introduced some restrictions in mid-January.
These include preventing Grok “from allowing the editing of images of real people in revealing clothing such as bikinis” and restricting the image creation feature of the AI chatbot to paid subscribers, X said in a statement.
The Commission stressed that the opening of a probe does not prejudge its outcome. (26 January)
For Eater, Bettina Makalintal reports on a refreshing trend in the restaurant industry: businesses embracing lo-fi, handwritten communications and announcements in response to the ceaseless AI slop we’re served across social media these days. From a grocery store’s hand-illustrated flyers to a burger joint’s Sharpie-scribbled menu specials, restaurants are choosing advertising with a human touch and rejecting polished AI-generated graphics.
It’s hard to think of a path of more friction than choosing to run a restaurant; why add even more friction to that? In these conditions, yes, it costs more to hire a photographer to promote your new menu than to generate one with AI, but it’s also increasingly clear that many people would rather see a badly taken, lo-fi photo than an engineered AI image. For one thing, it feels like a reflection of one’s values. To me, a restaurant that employs AI art is capitulating to the culture of slop, akin to reheating premade food from Sysco. It’s impossible to square with a value system that cares about craft and integrity.
For Men’s Health, Brian Underwood investigates the flood of sunscreen misinformation on Instagram and TikTok, where influencers push inaccurate data about chemical UV filters, play chemist and promote their own DIY formulas, or even encourage no sun protection at all. Underwood sorts fact from fear, talking with dermatologists, chemists, and professionals who are working to reshape the messaging around sun exposure and sunscreen.
Environmental concerns may have also shown up in your feed, with suggestions that these same filters are poisoning our oceans, giving rise to a whole category of sunscreens labeled “reef safe”—a meaningless marketing term that’s not regulated by the FDA (the same goes for products labeled “natural” or “organic”). Dr. Nelson’s point about “low-relevance models” is something to consider here too: The research demonstrating that oxybenzone, octinoxate, and other active ingredients may lead to reef bleaching was done in a lab setting where coral was exposed to levels of these chemicals significantly higher than has been found in the natural environment. It conveniently left out the fact that inorganic filters like zinc oxide (the one typically used in “reef safe” formulas) would have the same effect in similar doses.
A funny thing happened on the day OpenAI announced it was shutting down Sora, its video generation app: Iran went all in on synthetic propaganda and very quickly started winning the global meme war. The timing is a coincidence, no doubt, but it is the kind of coincidence that illuminates.
Watching the explosive virality of the clips offers a powerful lesson in asymmetric media operations. They deploy cultural sophistication, an understanding of online communities and the enormously powerful creation tools made available by American tech companies, tools that give everyone on the internet access to a personal reality distortion field — drones, but for your feed.
On Wednesday, as Donald Trump was trying desperately to talk down the oil markets with hints of a deal, a stream of videos, carefully calibrated for U.S., regional and third country audiences rolled out on X via embassy accounts, Russia Today, and disaffected Maga influencers. The clips, by broad social media consensus, are good. Some lean heavily on the extremely online grammar of the U.S. right. Some remix Hollywood characters and likenesses in exactly the way that OpenAI’s now nixed billion-dollar deal with Disney was supposed to sanction. Others lean more heavily into Islamic iconography, featuring Donald Trump and Benjamin Netanyahu as worshippers of Baal, the foreign demon god who figures in both the Quran and the Hebrew Bible. The Lego movie is an especially rich resource, but so are TikTok formats, and the kind of idealized AI figures beloved of Trump administration meme makers. You can watch a few of them here.
Notably, faked war footage is far from the dominant format. All of these clips foreground and celebrate their own artificiality: some are sentimental, some triumphal, many are full of the gleeful adolescent wit of gamers on discord forums.
Researchers have long been warning that generative tools will undercut the authority of visual evidence, compounding and accelerating the damage created by slower, cruder forms of fakery: photoshop, selective editing, even gaming clips passed off as combat footage. Of course, we are already there, and have been for a while. Russia has been the paramount master of this game, in Ukraine and in its ongoing influence operations around the world. But others have learned quickly. Last year, when India and Pakistan were engaged in a brief aerial battle, social media bullshit overwhelmed and compromised traditional coverage. More recently, Israel’s obliteration of Gaza was accompanied by a sustained and comprehensive blizzard of visually compelling misinformation, propaganda, and official lies.
That continues. On March 28, Israel killed three journalists in a targeted strike in Southern Lebanon, claiming without evidence that one of them, Ali Shoaib, was a member of Hezbollah’s Radwan forces. They later distributed a photograph of him in military fatigues to reinforce the point, but explained to Fox news that in fact, they’d had to photoshop the uniform in because no such picture existed.
Meanwhile, in the Trump administration’s domestic war on immigrants and political opponents, we’ve seen a complete resetting of norms around the tone of official communication and any expectation that it is rooted in fact. Nowhere was that more evident than in the altered footage posted by the White House of the arrest of the prominent Minneapolis activist Nekima Levy Armstrong in January. In the video, shared by the official White House handle, a handcuffed Levy Armstrong is sobbing, her skin visibly darkened. In fact, she had faced arrest calmly.
Questioned by reporters about this blatant falsification, deputy White House communications director Kaelan Dorr responded: "Enforcement of the law will continue. The memes will continue.” Collapsing the distinction between a meme and the factual record with the aid of AI is the final step in this administration's insistence that its preferred narrative simply is reality.
The problem for the White House and its allies is that their choices in tech policy, official communication, and press freedom level the playing field for information war in ways that Tehran’s media strategists understand and they, for all their immersion in online worlds, do not.
Iranian propagandists know that the currency of visual information online has already been completely debased. They’ve dealt with it plenty, and no doubt practiced it themselves in regional battles for narrative dominance. Their insight is that as cheap and easy as it is to create and distribute fakes, returns on the effort of mobilizing what disinformation researchers call “coordinated inauthentic action” are diminishing. They still do it, but it isn’t where the action is.
Sam Altman, Elon Musk and Mark Zuckerberg have, in a very practical sense, wrought this moment in concert with Peter Thiel, Alex Karp, JD Vance and Donald Trump. At their urging, the U.S. has surrendered unrivaled dominance in scarce, expensive information and cultural assets in exchange for a political economy of media that widely distributes cheap, abundant ones.
Tech leaders and conservative politicians have worked consistently for a decade to deprecate the trustworthiness of American journalism and constrain its liberties. They have smeared its practitioners as “enemies of the people”; they have captured the commanding heights of the broadcast and culture industries through crony deals, and they have launched an assault on both press freedom and standards, two assets that once made American news outlets the envy of the world. Needless to say, the economic collapse of traditional media companies fostered by Google’s and Meta’s advertising duopoly only served to deepen the damage. Jeff Bezos’s Washington Post shuttered its Middle East bureaus just days before the war began.
Meanwhile, lying from agency podiums and the Oval Office, makes Karoline Leavitt barely distinguishable from Baghdad Bob, Iraq’s minister of information in 2003 whose surreal, truth-dodging press conferences during the U.S.-led invasion made him a global laughingstock. And the DOGEing of both the nominally independent Voice of America, as well as the state department’s Global Engagement Center leaves the administration with neither broadcast nor digital counter-propaganda assets.
When no one can be trusted with the actual truth, we are left with the AI equivalent of 19th-century editorial cartoons, produced at industrial scale and distributed globally. America has little advantage in that war, particularly when it is at a moral, political and legal nadir.
If anything, Iran, which combines repression with an enormously rich literary culture, film scene and advertising market brings serious capabilities to the fight.
Of course, the ebbing of information power was already under way during the first Trump administration, and during Joe Biden’s term in ways that are indissociable from broader democratic decline. The “trust and safety” architecture adopted by big platform companies was designed — implicitly if not always visibly — to conserve information authority, and ensure that it functioned in broadly pro-democratic ways.
After the disastrous failures of the Rohingya genocide — which rights groups and UN investigators blamed Facebook for facilitating — and the fears surrounding the manipulation of the U.S. electoral environment in 2016, there was a clear threat to the commercial and political health of Twitter, Facebook and YouTube. Tech companies, governments, researchers and human rights experts devised rules and norms for content moderation grounded in existing standards, tools for detecting coordinated inauthentic behavior, and a framework for crisis response.
The community of practitioners and institutions that sprung up to combat the flesh-eating virus attacking the body politic were working with bandaids in the battlefield hospital even before Covid, a coordinated attack from the right, and the second Trump victory hit them, but they succeeded in imposing some limits. That project now lies in ruins.
The Stanford Information Laboratory has been shut down. Trust and Safety teams at Meta and X have been disbanded. The national security arm of the project, centered around the State Department is gone, and private funding for countering misinformation has largely dried up.
Where are the hyperscalers, the AI titans, whose tools are being so effectively deployed, in all of this?
The trust and safety people who do work at OpenAI are dutifully putting out reports every few months. They are detailing how they foiled efforts to use ChatGPT for a Chinese influence campaign aimed at Sanae Takaichi, the Japanese prime minister, and exposing a Russian content mill feeding African newspapers. “Pro-tip for governments,” wrote Head of National Security policy Sasha Baker on LinkedIn of the February report. “Please don’t use our products to spread lies online.”
Governments, in the world of Sam Altman’s “democratic AI” do not include that of the United States. OpenAI has not mentioned a single U.S. ally — let alone the administration itself — in these reports.
OpenAI has hired multiple ex-Clinton, Obama and Biden officials, and in their work a weird, attenuated piece of the old national security approach to information integrity lives on, alongside the project of selling products to the Pentagon. The company’s leaders clearly treat these issues as a complement to messaging around Western AI, or a picayune adjunct to the bigger questions of AI risk, which are handled way up in the organizational stratosphere, as they are at Anthropic.
Perhaps the larger lesson is that you can’t really shut down Sora, or put AI-generated video back in its box. If you choose to prosecute an illegal war of choice after surrendering the hard-won high ground of a robust, democratic information environment, high tech weaponry will not offset the deficit. On the contrary, you will have compounded the risk of both tactical failure and strategic geopolitical defeat. When that happens, and in some ways it already has, those who made this war, and their enablers in Silicon Valley, will have only themselves to blame.
Login
, X