On October 25, Brazil’s three-time and current president, Luiz Inácio Lula da Silva, surprised the entire world by overcoming an electoral setback in the first round (in which he was defeated by the far-right candidate Flávio Bolsonaro) and securing victory and the presidential seat in Brasília.

Lula, the legendary union leader during the dictatorship, secured 51.79%, equivalent to 64,256,232 votes, while Bolsonaro (son of the detained former president Jair Bolsonaro) received 48.21%, or 59,826,041 votes.

What is truly surprising about Bolsonaro’s defeat (given that most pollsters had predicted a very likely victory for him) is the enormous number of votes Lula managed to gain during the three weeks between the first and second rounds of voting. In that first round, Bolsonaro took first place with 48.21% (more than 56 million votes), and Lula came in second with 45.15% (just under 54 million votes).

Many believed that the nearly 9 million votes obtained by all the other candidates combined in the first round would be distributed in a way more favorable to Bolsonaro, since most of those votes were cast for right-wing and center-right candidates. This seemed even more evident after many of the leaders of those parties endorsed Bolsonaro.

Added to this question was the high probability that, in a society highly polarized between the two candidates, many voters who either cast blank or invalid ballots or abstained from voting in the first round would support one of the two candidates.

Indeed, both of these circumstances played out. According to available data, more than 14 million Brazilians who had voted for other candidates, cast invalid ballots, or abstained from voting participated in the runoff, supporting either Lula or Bolsonaro.

And this is precisely the key factor in understanding Lula’s victory. Assuming that very few voters would have switched their support from Lula to Bolsonaro or vice versa, it is important to note that of those 14 million votes that were redistributed, Lula managed to secure 10.37 million more votes than he received in the first round, while Bolsonaro managed to gain just under 4 million votes. This was enough for Lula not only to close the gap against his opponent but also to surpass him by a comfortable margin.

What happened in Brazil?

Faced with this surprising turn of events, hundreds of analysts in Latin America and around the world have rushed to present countless charts, hypotheses, and analyses in an effort to fully understand the reasons behind this victory. And the reasons have to do not only with the clearly impressive comeback by Lula’s Workers’ Party and its allies, but also with the fact that he appears to have put a halt to a seemingly unstoppable string of victories by the far right in Latin America.

Given this, the question seems clear: How did Lula manage to convince such a staggering number of people to vote for him in the runoff who had not done so in the first round? In response, technical reasons have been put forward (such as campaign strategy, the use of social media, and mass mobilization), as well as political ones (fear of a far-right government, social-class alliances, the formation of a broad-based coalition, etc.).

Among the technical reasons, Lula’s tremendous effort to turn the runoff campaign into a genuine, nearly constant mass mobilization (in both large cities and small towns) has been highlighted. The former union leader once again demonstrated his great charisma and ability to rally people before thousands of PT activists and members of other allied parties who fought day and night without stopping to overturn the result.

Added to this is the PT’s digital strategy; the party had repeatedly accused major social media platforms of boycotting the presidential campaign by hindering or reducing the visibility of Lula’s ads – but not those of Bolsonaro. In response, André Janones proposed a change in strategy.

Lula’s WhatsApp channel grew from 3.7 million subscribers to more than 10 million in the span of 12 days. Through this channel, users were instructed to post certain messages, search for information online, comment on specific posts, and so on. In other words, to circumvent an algorithmic system that, they claimed, was deliberately sabotaging Lula’s campaign.

The result was astonishing: despite the alleged media blockade on social media, the platforms were flooded with messages, images, and videos (many generated by AI) featuring Lula and attacking Bolsonaro’s candidacy, which undoubtedly had a major impact on the political communication strategy of the PT’s campaign, led by re-elected Governor Rafael Fonteles. For example, online interactions about Lula increased from 41% to 65% during the runoff election.

But Lula also took the risk of bringing about tangible changes in people’s lives, and he sent what may have been his final message during the runoff election, when his party and allies succeeded in passing legislation to end the six-day workweek. The move was celebrated by the country’s workers and clearly gave Lula’s campaign a pro-worker tone. That is why he gained so many millions of votes in the second round of the election in the working-class suburbs of major cities.

Read more: Brazil’s Congress ends 6×1 work schedule

And yet, there remains a certain suspicion that if this set of measures had been applied in any other similar situation, something would have been missing to set it in motion and give deeper meaning to the purely political campaign – with its communication, marketing, and other tactical decisions. And that something is Lula.

At times it seems that only someone with his political stature, his charisma, and the historic place he has already carved out in the history books could have won an election in Latin America against the far right, which enjoys the support of the most influential imperial powers of our time and an international oligarchy.

The “warrior of the Brazilian people

Lula achieved this, as already mentioned, thanks to a series of changes in his campaign strategy and concrete improvements in people’s lives. But it must be said without fear of contradiction: he did so by blending two dimensions that might seem contradictory, yet which a figure like him could bring together.

The first is that, as some of his campaign strategists recount, he began to speak extensively about Brazil’s future despite having already served as president three times. In this way, he moved away from nostalgia for the past and began to speak of hope for the future.

And second, Lula insisted throughout the campaign that this future was threatened by the Bolsonaros, whom he constantly linked to a submissive attitude toward US interests. In this way, Lula also employed a discourse of fear and danger, personifying it in Bolsonaro, who, as is well known, had reportedly left Brazil for the United States.

But Lula did not do so in a pitiful or apocalyptic way; rather, he presented himself as the only one capable of halting that advance. And more than just capable, as the only one who, without a doubt, would do so. In other words, Lula established himself, at the age of 80, as a “warrior of the Brazilian people,” just as his supporters chanted in the streets.

Then, in the debates (which undoubtedly had a huge impact on the runoff election) it seems Lula had had enough of the far right always appearing to be on the offensive, while the left and progressives always seemed to be waiting for the next blow. Lula put on his boxing gloves and began playing the game the far right had set for him … and won by knockout.

And for that very reason, Lula, his followers, and many of those who supported him (not necessarily because they believe in his agenda, but because they feared a new far-right government, as the numbers show) must realize that, with a hostile legislature and an opposition united around a single family, Lula’s fourth term will likely be the most difficult of his life.

Lula will have to seek ways to overcome the clear obstacles that the right and the far right will place in his path in order to achieve what he called “the best government in Brazil’s history.” For now, he has managed to prevent another Bolsonaro from reaching the presidency, but that does not mean the remainder of his term will be a “Brazilian pax.” 

In this regard, there is a Japanese proverb that Lula should keep in mind: “When you’ve won a battle, tighten your helmet.”

A new investigation into the activities of Google, Meta and other technology companies has intensified scrutiny of EU plans to revise data-protection legislation, placing Ireland and Germany at the centre of a dispute over corporate influence and European digital sovereignty.

European technology policy has become the focus of a dispute over corporate lobbying after researchers documented extensive contacts between major technology companies and political decision-makers during negotiations to revise EU data-protection rules.

An investigation published on 28 October by Corporate Europe Observatory and the German transparency organisation LobbyControl alleges that leading technology companies mounted a coordinated campaign to influence European governments and legislators.

The findings come one week after EU member states agreed a negotiating position on the Digital Omnibus, a legislative package intended to simplify aspects of European data-protection, privacy and cybersecurity regulation.

The campaigners argue that the proposed changes risk weakening existing safeguards in order to accommodate the commercial requirements of artificial intelligence developers.

Ireland, which holds the rotating presidency of the Council of the European Union, has rejected suggestions that its handling of the legislation was improperly influenced by technology companies.

The dispute exposes a fundamental disagreement over the balance between commercial innovation and the protection of personal information.

The scale of the lobbying campaign

According to LobbyControl’s investigation, non-European technology companies recorded 103 meetings with representatives of Germany’s federal government between May 2025 and September 2026.

Google accounted for 28 meetings, Microsoft 18, Amazon 16 and OpenAI 11.

Germany’s Digital Minister, Karsten Wildberger, participated in 18 meetings with representatives of major technology companies, the researchers reported.

In Ireland, the organisations identified extensive engagement involving Meta, Google and politicians ahead of the country’s assumption of the EU Council presidency in July.

Their findings indicate that Meta engaged with 41 Irish parliamentarians over the Digital Omnibus, while Google recorded 30 meetings.

The researchers also identified 655 lobbying meetings concerning the legislation involving members of the European Parliament.

According to their analysis, approximately 83 per cent represented business interests, compared with 11 per cent involving civil society organisations or trade unions.

These figures demonstrate the intensity of corporate engagement, although the existence of lobbying meetings does not, by itself, establish that particular legislative provisions were adopted because of that activity.

Why the Digital Omnibus matters

The legislative package forms part of a broader European Commission initiative to reduce administrative burdens and improve competitiveness.

Its provisions address several areas of digital regulation, including the General Data Protection Regulation, commonly known as GDPR.

The proposals seek to simplify compliance obligations, clarify the treatment of personal information and revise rules governing certain forms of data processing.

One particularly contentious issue concerns the use of personal data in developing and operating artificial intelligence systems.

AI developers require large quantities of information to train and improve their models. However, the collection and processing of personal information remain subject to European legal safeguards.

Technology companies have argued that regulatory complexity can discourage investment and make it more difficult for European businesses to compete internationally.

Privacy campaigners maintain that simplifying regulatory procedures must not become a mechanism for reducing substantive protections.

The disagreement concerns not merely administrative paperwork but the legal circumstances under which companies may process information relating to identifiable individuals.

Ireland rejects allegations of improper influence

Ireland occupies a particularly sensitive position in the debate.

The country hosts major European operations of several American technology companies, including Google and Meta, and has substantial economic interests in the digital sector.

At the same time, its presidency of the EU Council requires it to coordinate negotiations between member states.

According to RTÉ’s reporting on the investigation, the Irish government strongly rejected the allegation that it had weakened data safeguards in response to corporate pressure.

A presidency spokesperson emphasised that Ireland’s responsibility was to facilitate agreement between member states and that the negotiating mandate approved on 21 October enjoyed broad support.

The Council has presented the proposed changes as a means of reducing compliance costs while preserving established privacy protections.

In its official statement, the Council said its mandate retains the existing definition of personal data and introduces additional provisions governing pseudonymisation.

It also includes safeguards relating to sensitive information used for biometric verification and AI systems.

These details are significant because they demonstrate that the negotiating text contains protections alongside proposed simplifications.

Germany’s role in the negotiations

Germany has also emerged as an important participant in the controversy.

The investigation points to close contact between the German government and international technology companies during discussions about the future of European digital regulation.

Berlin has supported reducing administrative restrictions that it considers obstacles to innovation and the development of competitive AI technologies.

For Germany, the issue is connected to wider concerns about industrial competitiveness, investment and Europe’s dependence on foreign technology suppliers.

Yet an unresolved question remains.

Measures designed to assist European innovators may also benefit established American companies whose extensive computing infrastructure and data resources give them advantages over smaller competitors.

Consequently, easing restrictions does not necessarily guarantee a stronger European technology industry.

The European Parliament holds the next decision

The Council’s negotiating position does not make the Digital Omnibus law.

The European Parliament must establish its own position before negotiations between the institutions can determine the final legislation.

Parliamentarians will therefore have an opportunity to examine the treatment of personal data, the safeguards attached to AI development and the relationship between regulatory simplification and individual rights.

The lobbying investigation is likely to become part of that debate.

Its central political significance lies in the competing interpretations of European technological sovereignty.

For governments seeking investment, regulatory flexibility may provide an instrument for improving Europe’s international competitiveness.

For privacy organisations, the same flexibility could increase the influence of companies already dominant in global digital markets.

The outcome will help determine whether Europe’s future digital framework preserves the existing balance between commercial data use and individual protections, or establishes a different relationship between the two.

È in arrivo un’importante novità per Instagram, che si appresta a introdurre una funzionalità che consentirà di caricare video della durata massima di 90 minuti. La nuova funzionalità, che si chiamerà “Features“, potrà essere sfrutatta per portare sulla piattaforma di Meta tutti quei contenuti che al momento potevano essere fruiti solo al di fuori di essa. Molti podcast, ad esempio, potevano solo essere promossi tramite alcune brevi clip su Instagram, mentre ora intere puntate – purché la durata non superi l’ora e mezza – potranno essere caricate.

Per ora questa funzionalità è ancora in fase di sperimentazione: si sta, infatti, testando negli Stati Uniti, dove è disponibile per un piccolo gruppo di utenti, e con alcuni creator in Brasile, secondo quanto riferito da Megan Buckley, portavoce di Instagram. Ad annunciare questa novità è stato lo stesso capo di Instagram Adam Mosseri, che insieme Tessa Lyons, vicepresidente della sezione prodotti di Instagram, ne ha spiegato le ragioni.

“Reels rimarrà una parte fondamentale di Instagram. Ma quando si desidera approfondire la conoscenza dei creator preferiti e dei contenuti che creano, la possibilità di farlo dovrebbe essere disponibile su Instagram” ha spiegato Lyons. Mosseri ha anche raccontato di essere stato a lungo contrario sulla possibilità di introdurre su Instagram contenuti più lunghi dei venti minuti fino a oggi consentiti.

Ora il vero interrogativo rimane se questa operazione si rivelerà vincente. Al momento, infatti, Instagram non garantisce la possibilità di monetizzare in maniera diretta tramite i contenuti che si caricano sulla piattaforma, a differenza di quanto avviene ad esempio su YouTube dove i guadagni sono garantiti sia dalla pubblicità che dal numero di abbonati a un canale. Anche su Instagram esisterebbe la possibilità di abbonarsi al canale di un creator, ma al momento è un’ipotesi che viene quasi totalmente trascurata.

“È troppo presto per condividere un modello di monetizzazione a lungo termine, ma stiamo esplorando diverse modalità per consentire ai creatori di guadagnare da brand, fan e Instagram, inclusi incentivi basati sulle prestazioni” ha spiegato un portavoce di Instagram alla rivista TechCrunch.

L'articolo In arrivo i video da 90 minuti su Instagram: ecco cos’è e come funziona la nuova “Features” proviene da Il Fatto Quotidiano.

L’allarme lo ha lanciato, per prima, la Nieman Lab Foundation di Harward lo scorso 15 ottobre: “A partire dal 30 ottobre, gli account aziendali che continueranno a utilizzare Facebook, Instagram o WhatsApp acconsentiranno automaticamente alla raccolta di dati da parte di Meta sui propri siti web”.
Destinatari dell’allarme, innanzitutto, gli editori. Secondo la Fondazione Nieman, infatti, per effetto di un aggiornamento delle proprie condizioni generali di contratto business, Meta si sarebbe garantita il diritto di sfruttare i contenuti dei giornali per addestrare modelli e algoritmi alla base dei propri prodotti.

In poche ore, in tutto il mondo, gli editori si son messi sul piede di guerra. Tanto è bastato a suggerire a Meta di fare marcia indietro o, almeno, tirare il freno.

Il gigante di Menlo Park, infatti, ha rivisto e corretto l’aggiornamento delle condizioni generali di servizio all’origine della protesta degli editori, chiarito che non aveva mai avuto intenzione di garantirsi alcun diritto in più rispetto a quelli che già aveva e tanto meno di farlo contro la volontà degli editori e specificato che i contenuti degli editori, a valle della modifica in questione, avrebbero potuto essere utilizzati esclusivamente per migliorare i propri prodotti pubblicitari e commerciali, quelli, cioè, forniti proprio agli editori. E ha aggiunto che gli editori avrebbero, comunque, continuato a disporre del diritto di sottrarre qualsiasi contenuto al proprio scraping, semplicemente utilizzando l’apposito comando – il famoso robot.txt – da sempre a loro disposizione.

Meglio ma non bene secondo la stessa Fondazione Nieman che in un articolo del 17 ottobre è tornata sull’argomento, chiarendo, attraverso la vivavoce di una serie di esperti, che la modifica dell’aggiornamento alle condizioni generali di servizio proposta da Meta non sarebbe valsa a superare il problema, che la società sarebbe rimasta libera di utilizzare i contenuti degli editori per addestrare gli algoritmi alla base di una non meglio definita ampia gamma di “prodotti commerciali” e che, nonostante le rassicurazioni, niente nelle condizioni generali di servizio avrebbe, effettivamente, garantito agli editori di sottrarre, attraverso il comando robot.txt i propri contenuti alla bulimica voracità degli algoritmi di Menlo Park.

Nessun cessato allarme insomma. E, anche da questa parte dell’oceano, in tanti, tra gli editori, la pensano allo stesso modo.

Tra questi sicuramente gli editori norvegesi che, infatti, hanno appena trascinato Meta in Tribunale chiedendo al Giudice di ordinarle di sospendere l’aggiornamento dei propri termini d’uso, di ordinarle di inserirvi un esplicito riferimento alla circostanza che con il comando robot.txt sarebbe possibile sottrarre i propri contenuti all’addestramento di qualsivoglia prodotto e di imporre che qualsiasi questione relativa allo sfruttamento di contenuti editoriali sia disciplinata con un contratto completamente autonomo rispetto a quelli relativi ai servizi generalmente utilizzati per la diffusione e promozione dei contenuti editoriali, così da sottrarli al rischio di rimanere esposti a eventuali abusi di posizione dominante da parte della società di Menlo Park.

Ancora una volta la risposta di Meta non si è fatta attendere: aggiornamento dei termini d’uso posticipato dal 30 ottobre al 20 novembre per aver modo di spiegarne meglio agli editori il significato e di tranquillizzarli.
Ancora una volta, meglio ma non bene.

Il problema resta ed è epocale anche se, per la verità, non riguarda solo Meta ma tutte le grandi fabbriche globali di algoritmi: non è, infatti, tollerabile oltre l’idea che, specie soggetti che, in un modo o nell’altro, si sono ormai conquistati un ruolo ineliminabile nella catena di distribuzione dei contenuti informativi, provino a appropriarsi di questi contenuti, trasformandoli in asset tecno-commerciali attraverso i quali andare, addirittura, a fare concorrenza agli stessi editori dei contenuti in questione.

E non è solo una questione di soldi e di sopravvivenza delle imprese editoriali. È, anche, una questione di democrazia perché se pochi grandi giganti dell’intelligenza artificiale, divorando i contenuti dei giornali, si pongono in condizione di diventare – come, in parte, sta già accedendo – nuovi oligopolisti globali dell’informazione, la dieta mediatica dell’intera umanità finisce con il cadere nelle loro mani con ogni conseguenza facilmente immaginabile in termini di pluralismo informativo, manipolazione dell’opinione pubblica globale e, appunto, democrazia.

Ma c’è di più. Non solo, infatti, le grandi fabbriche di algoritmi non hanno alcun diritto di appropriarsi dei contenuti degli editori per addestrare i propri algoritmi senza neppure chiedere permesso ma – ed è bene dirlo subito e senza nessuna ambiguità – gli stessi editori non possono vendere ai giganti dell’intelligenza artificiale in questione i loro contenuti perché li usino per l’addestramento degli algoritmi.

La ragione è semplice: i contenuti in questione, infatti, non sono solo oggetto di diritti di proprietà intellettuale degli editori dei quali questi ultimi possono certamente disporre ma contengono anche milioni di dati personali di milioni di persone – i protagonisti dei fatti raccontati – sui quali, al contrario, gli stessi editori, non hanno alcuna forma di proprietà e dei quali non possono in alcun modo disporre.

Questi dati personali, infatti, sono presenti nei contenuti degli editori in forza di un’eccezione – quella che ne consente il trattamento in vista del soddisfacimento del diritto di cronaca – e non di un diritto disponibile degli editori e quell’eccezione, evidentemente, mentre consente la pubblicazione del contenuto, non consente all’editore, pure titolare dei diritti di proprietà intellettuale dei quali vorrebbe appropriarsi Meta e le sue concorrenti, di autorizzare chicchessia al loro sfruttamento al fine di addestrare, non importa quale algoritmo o modello, non importa in quale contesto.

Se vendessero alle grandi fabbriche di algoritmi i loro contenuti perché li usino per trasformarli in asset tecnologico-commerciale, gli editori farebbero esattamente ciò che, a ragione, stanno contestando ai giganti dell’intelligenza artificiale di fare: esproprierebbero milioni di persone di un loro diritto fondamentale e trasformerebbero, a loro volta, i dati personali di milioni di persone in un proprio asset commerciale, senza neppure chiedere permesso.

Quindi bene e, anzi, benissimo che gli editori, anche se in ritardo, si preoccupino di difendere la loro proprietà intellettuale, ma guai a dimenticarsi, che c’è un diritto, altrettanto prezioso, che va difeso con eguale energia: quello che milioni di persone, protagoniste, nei contesti più diversi, dei contenuti pubblicati dagli editori, hanno sui loro dati personali, dati legittimamente – fino a prova contraria e a condizione che esistesse un interesse pubblico – pubblicati sulle pagine dei giornali ma, non solo per questo, legittimamente cedibili alle fabbriche di algoritmi o da queste acquistabili per contratto.

E, ancora una volta, se accadesse sarebbe gravissimo perché, quei dati, i nostri dati personali, una volta finiti nei ventri molli e profondi delle fabbriche di algoritmi, sarebbero sottratti per sempre al nostro controllo e, con loro, lo sarebbero frammenti di inestimabile valore della nostra identità personale.

L'articolo Gli editori norvegesi portano Meta in tribunale: il problema è epocale proviene da Il Fatto Quotidiano.

• Have an AI agent? Need a hammer? Home Depot is thinking about how to make that palatable for the retailer.
• Angie Brown, the company's CIO, spoke to Business Insider about agents, the PHP cyber scramble, and AI fixes.
• She also talked about Home Depot's involvement in Project Glasswing, which grants access to Anthropic's Mythos Preview.

Home Depot wants to roll out the welcome mat for your AI agent — so long as it's wearing a name tag and acting on your behalf.

During a sit-down interview with Business Insider, Angie Brown, the company's executive vice president and chief information officer, said the retailer wants to support AI agents that can search and buy products on a customer's behalf. But Home Depot needs to be sure of two things first.

"We want to know you're not a bot, and we actually want to know if it's you," she said. "That's something that Muse is working on right now, to self-identify."

Her comments are the latest glimpse into how legacy retailers — and their tech staff — are thinking about the possible next generation of online shopping.

Recently, several Silicon Valley companies have launched consumer-focused AI agents, such as Meta's Muse and Spear Street Technology's Instinct AI. They promise agents capable of scanning the internet, comparing products, filling carts, and buying items with little to no intervention from the person who sent them.

Their rollout has been met with varying levels of resistance and buy-in from online retailers. Amazon barred Muse from shopping on its marketplace, saying the bots didn't properly identify themselves. Several other retailers — including Walmart, Best Buy, Sephora, and Shopify — have announced integrations or openness to AI agents. In October, Meta announced that it co-developed a personal agent protocol with Sierra AI to standardize AI agent identification.

Brown said Home Depot is open to the idea of AI agents, but the retailer doesn't want to lose its connection with loyal customers.

Frequent Home Depot shoppers, such as contractors and remodelers, are part of the company's Pro Rewards platform, where they're eligible for discounts. The retailer doesn't "want to lose contact with the customer," she said.

"That's the next evolution," Brown added, confirming that Home Depot is in active conversations with Meta about how Muse should introduce itself to websites. "There's openness on both sides, because they want it to grow."

Home Depot is already using AI to defend itself

Brown said Home Depot is deploying AI amid a recent wave of cybersecurity worries.

Last week, companies around the world raced to patch critical vulnerabilities in PHP, a programming language used in an estimated 70% of websites. The flaws were discovered with help from OpenAI and prompted US cybersecurity officials to urge affected organizations to update their systems or, in some cases, take them offline.

Brown, whose remit includes Home Depot's cybersecurity, called the hours after the warning a "little hectic," but said the company's main website was not affected.

"We had a couple of operating units — companies we've bought and things like that that are a little bit smaller — that had to plug a few vulnerabilities," she said, including an internal job application system that was taken down "in the middle of the day for about an hour to patch."

Otherwise, she said, the company was largely spared from the sudden tech anxieties.

Home Depot, Brown confirmed to Business Insider, has joined Anthropic's Project Glasswing, an initiative that gives selected companies access to Claude Mythos Preview to help secure critical software. She said the company was not granted access in the project's first wave, "despite our best efforts, and calling a lot of friends." Those slots were largely reserved for banks and cybersecurity firms, not major retailers.

The Anthropic models "found a few things" for Home Depot to fix in its code — but the agents also found a few "false positives," Brown said.

Home Depot, she said, has long ranked cybersecurity vulnerabilities, but AI is leading the company to rethink its desired time to deploy a fix.

"The speed to fix a vulnerability has to be driven down to hours," she said.

Brazil’s Supreme Federal Court ordered an investigation Tuesday into Flávio Bolsonaro’s presidential campaign over suspected illegal foreign financing, algorithmic manipulation on digital platforms and possible deals involving strategic minerals.

The ruling targets the campaign of the far-right senator who lost the 2026 runoff to Luiz Inácio Lula da Silva. Justice Flávio Dino issued the decision under a constitutional action (the Fundamental Precept Non-compliance Argument 1178 (ADPF), after formal requests from Lula’s campaign and from former presidential candidate Renan Santos of the Missão movement.

The Court stressed that Brazil’s Constitution and laws expressly bar parties and candidates from receiving funds from foreign entities or governments.

RELATED: Flávio Bolsonaro to Face International Detention Request and Interpol Alert

At the center of the inquiry sits Rockbridge Network, a conservative U.S. organization tied to the political origins of JD Vance, now U.S. Vice-president.

Federal Police will determine whether the group, through the alleged mediation of Brazilian businessmen Tallis Gomes and Pedro Sang, offered financial resources, services or economic advantages to the campaign of the Senator Flávio Bolsonaro, the Liberal Party’s presidential candidate.

Rockbridge Under Scrutiny

Investigators will also examine possible underlying geopolitical and economic interests. The probe will look into whether negotiations linked political financing to the exploitation and control of strategic minerals and rare earth elements on Brazilian soil, resources that have drawn marked interest from the government of Donald Trump.

The Court framed the case as a matter of national sovereignty. Foreign interference in an election, the ruling argued, does not merely distort the vote; it compromises the state’s ability to decide its own direction. That framing elevates the investigation beyond a routine campaign finance dispute.

Ten federal agencies received the order to carry out the work, which includes the Brazilian Intelligence Agency (Abin), the Council for Financial Activities Control (Coaf), the Federal Revenue Service and the Central Bank.

Meta Faces Investigation

The technological front carries equal weight. The Supreme Court ordered an inquiry into the role of Meta, owner of WhatsApp, Instagram and Facebook, after detecting what the ruling describes as “intentional asymmetries and algorithmic manipulations.”

Case documents reveal internal messages in which employees admitted restrictions on the promotion of Lula’s ads while Flávio Bolsonaro’s profile received irregular positioning. UN experts had previously warned of the risk of an artificial electoral imbalance.

Federal Police will also trace paid content favorable to the far right that was distributed through accounts based abroad without proper identification of the contractors, as well as the use of Polymarket projections to tilt public perception.

The pressure on Meta is not new. On October 15 a judge had already barred the company from interfering in the distribution of electoral content following complaints from progressive sectors. Meta denied any favoritism, attributing the imbalance to an artificial intelligence glitch that benefited the Liberal Party candidate. The current process now escalates to criminal and national security levels.

For Flávio Bolsonaro, the investigation compounds an already difficult legal position. The same Court recently ordered him to surrender his passports, and lawmakers have sought his preventive detention and an Interpol red notice over a reported flight to United States. The new probe broadens the scope from his personal conduct to the machinery that financed and amplified his campaign.

Whether Rockbridge or Meta ultimately face charges depends on the evidence Federal Police gathers. The Court’s message, however, is already clear: foreign money and foreign platforms will not operate in Brazil’s elections without scrutiny.

Uma carta assinada por 13 parlamentares democratas dos Estados Unidos cobra explicações de Mark Zuckerberg, CEO da Meta, sobre denúncias de tratamento desigual às campanhas de Luiz Inácio Lula da Silva (PT) e Flávio Bolsonaro (PL) nas plataformas da empresa durante as eleições presidenciais brasileiras de 2026. Enviada em 22 de outubro, a carta ocorre em meio a denúncias de restrições a conteúdos políticos, redução de alcance de publicações e diferenças no funcionamento de ferramentas de inteligência artificial, episódios que levantaram suspeitas sobre a atuação de empresas estrangeiras no processo eleitoral brasileiro.

Entre os signatários estão o senador Bernie Sanders e os deputados Pramila Jayapal, Greg Casar, Maxwell Frost, Jesús G. (conhecido como Chuy García), Jonathan L. Jackson, Summer L. Lee, Mark Pocan, Mike Quigley, Delia C. Ramirez, Jan Schakowsky, Rashida Tlaib e Nydia M. Velázquez. 

Os parlamentares afirmam estar preocupados com os diversos relatos que surgiram ao longo das eleições brasileiras de que as plataformas da Meta estariam restringindo ou reduzindo a distribuição de conteúdos ligados à campanha de Lula, enquanto ampliavam ou recomendavam publicações associadas a Flávio Bolsonaro e seus aliados. Segundo os signatários, as diferenças de tratamento poderiam criar condições desiguais na disputa eleitoral.

Entre os episódios citados está a desativação, em 23 de setembro, da conta de anúncios da campanha de Lula no Facebook, 11 dias antes do primeiro turno. Embora a conta tenha sido restabelecida após 24 horas, os parlamentares afirmam que a Meta não apresentou uma justificativa detalhada para a medida. A carta também menciona restrições ao impulsionamento de anúncios da campanha no Instagram entre 9 e 10 de outubro, que, segundo uma estimativa da própria campanha, teriam impedido cerca de 1,3 milhão de pessoas de visualizar as publicações.

O documento cita ainda uma reportagem da Folha de S.Paulo sobre comunicações internas da Meta que indicariam que funcionários da empresa tinham conhecimento de recomendações algorítmicas que promoviam o perfil de Flávio Bolsonaro e um canal do WhatsApp associado à sua campanha, prática que é proibida pelo Tribunal Superior Eleitoral (TSE).

A carta também menciona um levantamento da organização Ctrl Z, divulgado pela Agência Pública, que documentou mais de mil casos de restrições e banimentos nas plataformas da Meta na primeira quinzena do mês. Entre os conteúdos cuja orientação política pôde ser identificada, nove em cada dez eram favoráveis a Lula ou progressistas. O levantamento registrou casos de perda de alcance, restrições às funcionalidades de contas e remoção de publicações.

Os parlamentares dos Estados Unidos pedem que Zuckerberg esclareça quais medidas foram adotadas, quais regras orientam a moderação de conteúdo político em eleições estrangeiras e se a empresa mantém comunicações com integrantes do governo Donald Trump sobre o Judiciário brasileiro, a regulamentação das plataformas e as eleições de 2026.

O documento também questiona como a Meta pretende garantir que restrições automatizadas às contas de anúncios dos candidatos sejam analisadas por pessoas em até duas horas. Por fim, os signatários perguntam se a empresa permitirá que um auditor independente examine dados internos de moderação e distribuição de conteúdo político relacionado às eleições brasileiras.

Os parlamentares estabeleceram o dia 4 de novembro como prazo para o envio das respostas por escrito, sem retorno até o momento de publicação dessa nota. 

Alertas da sociedade civil 

Para Luã Cruz, diretor de Litigância Estratégica da Ctrl+Z, os episódios reforçam os alertas sobre os riscos de concentrar o debate político nas plataformas controladas pelas Big Techs. “A sociedade civil brasileira de direitos digitais tem alertado há anos sobre o problema de deixar todo o sistema de discussão política pública dentro de ferramentas que são controladas por empresas estrangeiras. As eleições mostraram isso justamente por ser um momento delicado”, afirma.

Segundo Luã, as denúncias reuniram evidências de diferentes formas de interferência, desde anúncios com desinformação e pagamentos em moeda estrangeira até o descumprimento de regras eleitorais e a derrubada de conteúdos progressistas: “Isso reforça de maneira mais aguda que a gente não pode depender delas”, diz.

O diretor ressalta que as plataformas continuam sendo importantes para a comunicação política – “Não é à toa que o canal dos Janones ajudou incrivelmente a campanha do Lula” – mas defende que o país desenvolva alternativas para reduzir a dependência desses serviços.

Outro episódio destacado é o funcionamento da Meta AI, assistente de Inteligência Artificial da Meta no WhatsApp. O Núcleo Jornalismo reproduziu testes em que a ferramenta se recusava a criar imagens com referências a Lula, ao PT e ao número 13, mas atendia a comandos equivalentes relacionados a Flávio Bolsonaro, ao PL e ao número 22. A Meta reconheceu a falha e afirmou ter adotado medidas adicionais para impedir a geração de imagens de candidatos, inclusive quando os comandos usavam apenas o primeiro nome e o número eleitoral. A organização Ctrl+Z também apresentou uma denúncia formal à Corregedoria da Justiça Eleitoral sobre a diferença de tratamento. 

A carta dos parlamentares norte-americanos também menciona a ocorrência, com base em reportagem publicada por Mônica Bergamo em 14 de outubro, que apontou diferenças na geração de imagens favoráveis aos dois candidatos. As denúncias levaram o TSE a determinar que a Meta garantisse tratamento igualitário às candidaturas. 

Na decisão, porém, o ministro Marco Aurélio Bellizze afirmou que, embora houvesse indícios de falha na ferramenta, não havia elementos suficientes para concluir que ela decorreu de um critério deliberado ou que ocorreu de forma sistemática. 

Ministério Público acionou a Justiça

As denúncias também motivaram medidas do Ministério Público Federal (MPF). A Agência Pública relatou que o órgão adotou iniciativas para tentar impedir bloqueios de contas e restrições de alcance considerados indevidos nas plataformas da Meta.

Na Justiça Federal do Rio Grande do Sul, o MPF ajuizou uma ação civil pública com pedido de urgência para impedir decisões de bloqueio de conteúdo baseadas exclusivamente em algoritmos e exigir que reavaliações e recursos sejam analisados por equipes humanas especializadas. O órgão também pediu a possibilidade de aplicação de multa diária de R$ 500 mil por conta afetada em caso de descumprimento.

Em outra frente, a Procuradoria Federal dos Direitos do Cidadão (PFDC) encaminhou uma recomendação extrajudicial à Meta, cobrando providências de transparência e proteção à liberdade de expressão durante o período eleitoral. A empresa recebeu prazo de 48 horas para informar se acataria as recomendações e de 30 dias para apresentar um relatório conclusivo.

As medidas foram motivadas por denúncias apresentadas pelas organizações Ctrl+Z, Artigo 19, Conectas e Sleeping Giants Brasil. Entre os casos relatados estão suspensões de contas e restrições a conteúdos políticos favoráveis a Lula e contrários a Flávio Bolsonaro. A apuração também identificou casos de perfis e publicações removidos sob acusações de terrorismo e nudez, sem que houvesse conteúdo que justificasse essas classificações.

O MPF considerou genéricas as respostas apresentadas pela Meta e apontou que a empresa não esclareceu individualmente os motivos das restrições nem detalhou todos os casos de suspensões e remoções questionados.

Na avaliação de Luã, a reparação dos usuários afetados deve ser uma das prioridades na responsabilização da empresa. O diretor também defende a abertura de dados sobre a moderação de conteúdo e a aplicação de multas proporcionais à dimensão das plataformas.

“Além de você reparar o dano para o consumidor como um todo, tem que ter a abertura dos dados para a gente entender o tamanho do estrago. E, a partir disso, ver o terceiro ponto que é a multa. Tem que multar pesado. Essas empresas são as mais ricas da história do capitalismo e elas precisam ser responsabilizadas à altura”, diz.

Comitê de Supervisão aponta falhas de moderação

O Comitê de Supervisão da Meta também apontou falhas na aplicação das regras de moderação durante as eleições brasileiras. O colegiado criticou as restrições técnicas impostas às contas de anúncios de Lula e considerou que houve aplicação excessivamente rigorosa das políticas antifraude da empresa.

O comitê avaliou que o caso deveria ter sido encaminhado a uma equipe interna de especialistas, além de destacar que erros de moderação durante períodos eleitorais podem ocorrer tanto pelo excesso de rigor, que restringe manifestações políticas legítimas, quanto pela aplicação insuficiente das regras, que pode comprometer a integridade eleitoral. E cobrou mais transparência e consistência nos procedimentos de moderação e recomendou a criação de indicadores públicos que permitam avaliar a eficácia das medidas adotadas pela empresa, incluindo suas principais falhas.

A manifestação retoma preocupações sobre a atuação da Meta em eleições anteriores em outros países, inclusive nos EUA, e a necessidade de mecanismos que permitam avaliar os resultados de suas políticas de moderação. Embora a empresa divulgue dados gerais sobre anúncios e contas removidas, o comitê apontou limitações nas métricas públicas disponíveis para medir a efetividade dessas políticas.

TSE cobrou tratamento igualitário

O TSE determinou, em 15 de outubro, que a Meta oferecesse tratamento igualitário aos dois candidatos. Em 19 de outubro, diante do que o tribunal considerou descumprimento da decisão, a multa aplicada à empresa foi dobrada.

No dia seguinte, a Meta declarou ter concluído a implementação de 17 iniciativas para cumprir a determinação, incluindo um mecanismo de revisão humana das contas de anúncios das campanhas presidenciais.

Os episódios também evidenciaram limitações na atuação das instituições brasileiras diante das plataformas, aponta Luã. “Não foi uma questão de falta de instrumento jurídico ou até mesmo falta de conhecimento, por exemplo, da sociedade civil ou da população brasileira sobre o que estava acontecendo. Faltou ao TSE atuar de maneira diligente”, afirma.

Segundo o diretor, a sociedade civil alertava para os problemas desde março, quando foram publicadas as resoluções eleitorais sobre tecnologias. 

A Nvidia e a Meta anunciaram nesta terça-feira (27) que ​irão aprofundar sua parceria ​em torno do assistente de IA Muse e de modelos de IA relacionados, com a Nvidia investindo US$3 bilhões na oferta de ações da controladora do Facebook.

Os modelos e agentes ⁠Muse da Meta, ​que rodam na infraestrutura da Nvidia, ​trabalharão em colaboração mais estreita com a ⁠fabricante de chips para melhorar ⁠o desempenho e reduzir custos.

O ​Muse, ‌conhecido internamente como Hatch, é fundamental para ⁠o plano do presidente-executivo Mark Zuckerberg de oferecer “superinteligência pessoal” aos bilhões de pessoas que utilizam os ‌serviços da ⁠Meta diariamente.

Inspirado ‌no agente de IA de código aberto OpenClaw, o Muse pode acessar emails, calendário, ⁠pagamentos, informações de saúde, ⁠compras e aplicativos de casa inteligente do usuário, sendo ‌que os usuários podem revogar o acesso a qualquer momento.

A Meta também aumentou sua oferta de ações para US$ 97,5 bilhões, a maior ‌captação de recursos de sua história, à medida que acelera os investimentos em infraestrutura de ⁠IA.

As empresas podem explorar uma colaboração mais estreita na estrutura OpenShell da Nvidia, incluindo o ​uso dessa estrutura para implantar software com segurança ​e o uso de uma ferramenta chamada Sentry para monitorar o comportamento dos agentes de IA.

Meta lança IA que faz compra pelo usuário | ABERTURA DE MERCADO

• Instagram is opening its arms to long-form video again.
• Adam Mosseri, Instagram's top exec, said longer videos will help the app stay "relevant."
• Instagram is testing tools for creators making episodic content. But will it pay for them?

Instagram's top executive, Adam Mosseri, has a plan to keep the platform "culturally relevant" — by making it more like YouTube.

Meta-owned Instagram unveiled an overhaul of long-form video features on Monday, including a test of horizontal videos called "Features." Taken together, the changes make the app feel a lot more like YouTube.

"We want to be the most culturally relevant platform," Mosseri said in an interview with content creators and founders of Press Publish, Colin Rosenblum and Samir Chaudry. "To do that, we just need to show up where people are."

The context: YouTube has become a major hit on TV screens, and there's a growing appetite for Instagram-native shows with recurring episodes.

"Subway Takes," hosted by Kareem Rahma, is a prime example of Instagram's strengths and limits. The show started in 2023 as a short-form interview series and, as it grew, expanded with longer-form cuts on YouTube and the streaming platform Tubi. Instagram, however, is where Rahma's show still has its largest audience of 2.4 million followers.

Instagram doesn't want people to go to YouTube or Tubi to see longer versions of shows like "Subway Takes." That's leaving ad revenue and user engagement on the table. Its new tools are designed to entice creators and viewers to stay on its platform.

Here are Instagram's latest video tools and how creators can use them:

• "Features" keep viewers within the Instagram ecosystem. Instead of leaving Instagram to watch the extended version of an interview or episode, it'll be relatively seamless to stay on Instagram and watch that longer-form content. Instagram's "Features" can be as long as 90 minutes, although the new format is, for now, only available to a limited number of creators in the US and Brazil.
• Instagram is feeding more content to its TV apps. Instagram is also pushing its "Instagram for TV" app further. The app will feature various video content on TV services like Samsung, Amazon Fire TV, Google TV, and now Apple TV. People can now watch a feed of content their friends are also liking or resharing on TV.
• Instagram-native shows can label their episodes more clearly with "Series." Within the main Instagram feed, Instagram is testing a way for shows on the app to seamlessly connect episodic content. Eligible creators can add videos to a Series, and a label appears over the content indicating that there are additional episodes. People can then watch those episodes in order from the start. This test is expanding to more countries through the end of the year.
• Clips can directly drive engagement back to the original video. Instead of clipping content and praying that the audience will go the extra mile of clicking on an account and finding the original, longer-form video, Instagram is letting creators link up to 8 clips to one long video.

This isn't Instagram's first time dabbling in long-form content.

In 2018, it launched IGTV, a long-form video format and stand-alone app. In 2020, Instagram rolled out a program that let creators earn money from those videos on IGTV. Four years after launching, the company sunset the IGTV app and shifted its video efforts to Reels.

There are a few key elements missing from Instagram's big leap (back) into long-form video.

You can't watch the video while continuing to scroll through Instagram, or multitask by opening other apps while keeping the Instagram content in a picture-in-picture pop-up. There also aren't automated transcripts and subtitles yet. A spokesperson for Meta told Business Insider that there are plans for these features.

Will creators get a slice of the ad revenue pie?

For creators, one key question and pain point looms: Will Meta share the honeypot?

Mosseri told Rosenblum and Chaudry that Instagram is thinking about ways to test pre-roll and mid-roll ads that could lead to a revenue split. However, he said previous attempts by Instagram to build a creator monetization system comparable to YouTube's Partner Program didn't prove beneficial for the platform.

"We spend more money than we make back by a long shot," Mosseri said of Instagram's previous efforts at broad revenue-sharing programs.

Mosseri said that in its previous iterations of content monetization, creators typically didn't produce more content than they already did for free, and the content they did create tended to be lower quality.

Still, Instagram will need to "figure it out" as it tries to crack longer-form content, Mosseri said.

"I don't think of platform payments or performance-based incentives or bonuses as a revenue driver," Mosseri added. "I'd be psyched to scale them up for any of these formats — as long as we're breaking even."

Meta Platforms revealed a plan to secure up to $95 billion in equity to fund its AI-related expansion ​of data centres and cloud computing, backed by chipmakers Nvidia, AMD and Broadcom.

The social media company aims to raise $35 billion in concurrent underwritten public equity offerings and up to $60 billion as part of an at-the-market programme scheduled to begin in Q1 2027.

Meta Platforms stated the chipmakers expressed an interest in becoming cornerstone investors through the equity offering.

They can purchase a total of $5 billion shares of Class A common stock in the offering, though Meta Platforms emphasised their expressions “of interest are not binding agreements or commitments to purchase”.

Meta Platforms stated AI is “creating significant opportunities” and its investments to-date are “generating tangible returns through increased engagement, improved advertising performance and increased monetisation”.

Last week, the company reported net income rose from $2.7 billion in Q3 2025 to $9.9 billion in the recent period, with revenue up 26% to $64.6 billion.

The post Meta seeks $95B to fund AI ambitions appeared first on Mobile World Live.

2016 was a pivotal year in the 21st century. In June, Britons voted in a referendum to leave the European Union. In November, U.S. voters elected Donald Trump as president, a figure widely regarded as a bad joke by both Democrats and establishment Republicans. Both events marked the beginning of a far-right wave that, a decade later, continues to sweep across much of the world. And both were aided, perhaps decisively, by a small London-based political consultancy: Cambridge Analytica. The company collected data from 87 million Facebook users without their consent and used it to create personalized political propaganda to boost the campaigns for Trump and Brexit, among others.

Seguir leyendo

• Meta is seeking a major cash injection to fund its AI ambitions.
• The tech giant said on Monday it is looking to raise up to $95 billion from stock offerings.
• Tech firms are spending big on AI infrastructure as they try to win the superintelligence race.

Meta is tapping investors for up to $95 billion to fund its huge AI build-out.

The tech giant said in a Monday filing that it has launched about $35 billion in equity offerings, comprising around $22.5 billion of Class A common stock, and $12.5 billion in depositary shares representing preferred stock.

The parent company of Facebook, WhatsApp, and Instagram also said it was planning a $60 billion at-the-market offering once its Class C shares begin trading in the first quarter of 2027.

In an at-the-market offering, a company sells shares gradually into the market at the prevailing market price, instead of as a large block at a fixed price.

Meta said last week that it would introduce the new share class to help founder and CEO Mark Zuckerberg access the liquidity he may need to pay California's proposed wealth tax without having to sell existing shares and erode his voting control over the company.

"AI is creating significant opportunities for Meta across its core business and new products and services," the company said in a press release.

"Meta is investing to expand the infrastructure and compute capacity needed to support these opportunities and to bring personal superintelligence to billions of people and businesses," it added.

Nvidia, AMD, and Broadcom have each expressed an interest in buying up to $5 billion of the common stock being offered, but those indications aren't binding, so they might buy more or less, Meta said.

Meta stock swung during Monday's premarket from 2% in the green to 0.4% in the red.

Meta is spending heavily on AI infrastructure as it develops consumer AI products and pursues Mark Zuckerberg’s goal of "personal superintelligence."

The company said it expects to incur $135 billion to $145 billion of capital expenditures this year. It highlighted $97 billion of operating cash flow for the first nine months of this year, and the roughly $82 billion of cash and other liquid assets it held at the end of September.

Meta's revenue jumped by 26% to around $65 billion last quarter, but its operating income shrank by 37% to $13 billion as costs surged from about $31 billion to $52 billion.

• Big Law has plenty of competition for its best lawyers.
• Some of the most ambitious lawyers want to work for clients in their legal departments.
• New Pirical data shows which law firms have the most alumni working in Big Tech.

Big Law has a hard time keeping its lawyers.

It's not just rival firms dangling basketball-player salaries, extra vacation, and branded merch to lure away top talent.

Big Tech is competing for those lawyers, too.

New data shared with Business Insider reveals which firms send the most lawyers to the biggest names in tech.

Pirical, a legal people-analytics company, tracked lawyers who previously worked at the 200 largest law firms by revenue in the United States and now hold legal roles at seven tech companies: Google, Meta, Microsoft, Nvidia, OpenAI, Anthropic, and SpaceX. Pirical filtered out summer associates and interns.

Morrison Foerster and Wilson Sonsini tied for the top spot, each with 40 alumni working in legal roles across the seven companies. Both firms regularly represent tech startups and the investors backing them.

Latham & Watkins followed with 33 former lawyers. Kirkland & Ellis, the industry's revenue leader, ranked 11th with 21 former lawyers working at those companies, tied with Paul Hastings and Sidley Austin.

Pirical's analysis may have missed lawyers whose job titles fell outside its search criteria, said Craig Savitzky, its insights and content editor. He said newer companies, particularly Anthropic, are also likely undercounted because their legal teams are growing faster than public profiles are updated.

Big Law-Big Tech pipeline

For attorneys hoping to land a coveted job at a tech giant, a stint at a white-shoe law firm can be a stepping stone. They can trade billable-hour quotas for a seat closer to the business, attractive pay, and just one client to keep happy instead of dozens.

That combination makes these roles "some of the most highly desirable and sought-after positions" for law firm associates, said Casey Sullivan, a recruiter with CMW Legal Search.

There's another draw, Sullivan said. The companies at the forefront of artificial intelligence are wrestling with some of the thorniest legal questions of the day, from intellectual property to regulation.

"That's where the most interesting work is for lawyers," he said.

Not all law firms are equally effective launch pads. Pirical's Savitzky noted that the firms near the top of the list have deep ties to the tech industry, if not the deepest pockets.

Cooley brought in $2.4 billion in revenue last fiscal year, placing it 29th on The American Lawyer's ranking of top law firms by revenue. But it tied fourth on Pirical's list, with 29 alumni working at the seven tech companies, including nine at Nvidia, a longtime client.

Davis Wright Tremaine, which has represented Microsoft in antitrust and employment lawsuits, has 13 alumni working in the software giant's legal department. Fenwick, which advised Wiz and Fitbit on their sales to Google, has 13 lawyers at the search giant.

Business Insider also sliced the data another way, looking at which of the seven tech giants employ the most alumni from the top 25 firms on Pirical's list.

Big Law is effectively a feeder school for Google's legal team. The company employs 322 alumni of the top 25 law firms on Pirical's list, more than any other tech company studied.

That doesn't necessarily mean Google is the most desirable location for lawyers leaving Big Law. Savitzky said the numbers likely reflect the sheer size of Google's legal department and the decades it has had to recruit from top firms.

• Canva launched Canva Craft, a tool similar to AI image models by Google, Meta, and OpenAI.
• Surprisingly, Canva fell short on a design task but did well at generating more realistic images.
• Here is how Canva's new image generation tool stacks up against ChatGPT, Gemini, and Muse.

Canva wants a shot at winning the AI image generation race.

The design platform launched Canva Craft on Thursday, its first image-generation model, similar to those from Google, Meta, and OpenAI.

Users can access Craft through Canva’s Dream Lab, including with free accounts, though it is subject to AI usage limits. You’ll need to sign in to a Canva account to get started.

In the release announcement, Canva said the tool plans a design’s layout, proportions, and text before generating an image, an approach that aims to help it produce legible lettering and accurate details.

The company also said that the model was trained on human-designed Canva templates, including posters, menus, and presentations, rather than photographs.

To test how Canva performs, Business Insider created three image generation prompts with escalating difficulty levels — from designing a simple greeting card to creating a photorealistic image from scratch.

Here is how Canva's brand-new image generation tool stacked up against three of the most powerful models.

Canva did not respond to a request for comment.

Starting off with an illustration

We started off simple with a prompt for a birthday card.

Create a cartoon "Happy Birthday" greeting card for a friend named Jessica who has pink hair, brown eyes, likes to wear paisley, and is turning 28 years old in November.

Surprisingly, Canva, the design platform, fell short in its design of the birthday card, from the awkward font to the outsize head of the cartoon character. It didn't even bother to include any paisley.

By comparison, ChatGPT and Muse understood the assignment.

ChatGPT guessed that the recipient's favorite color might be pink, while Muse thoughtfully wove in cozy fall elements for the November birthday girl.

Meanwhile, Gemini took the request literally and produced an image of an actual card. Extra points!

Winner: Muse

Editing an image

For the second task, I tested the AI models' ability to identify an element and place it in a brand-new setting.

Edit this image of me holding a necklace pendant. Take the red garnet pendant off my hand and make it look as if it is displayed in a vitrine in a jewelry showroom.

Canva did a great job of identifying the pendant and creating an atmospheric setting, but, like Gemini, it depicted the jewelry in a way that defied physics and common sense. It would be unrealistic for a pendant to float in the air or sit propped on a metal pole.

ChatGPT and Muse once again outperformed other models in understanding how things typically look in reality. Muse, however, made some alterations to the pendant, which I didn't ask for.

Winner: ChatGPT

Creating a photorealistic image from scratch

With Halloween approaching, I asked for a realistic depiction of my favorite animal.

Generate a photorealistic image of a long-haired orange tabby cat in a Halloween costume with bat wings, perched on the edge of a gothic medieval castle under the full moon and a starless night sky.

Though the cat's fur looks a little too strandy and high-contrast in Canva's creation, no components are missing or malformed, and the image composition pretty much matches the quality of Muse and Gemini.

The clear winner for me in this round, however, would be ChatGPT.

Not only is the cat's posture majestic, but the soft glow on his left side is extremely realistic, and the white balance is well done. There are also clouds softly cradling the moon, which could be a response to my request for a starless sky.

Winner: ChatGPT

Overall, it's evident that Canva still has a lot of room to grow compared to other models, especially if graphic design is its focus. For now, I'll stick to moving components around by hand.