Demonstrators holding a banner that reads in Portuguese "Brazil decides about Brazil" rally against Meta over allegations of interference in Brazil's presidential election, outside Meta's headquarters in Sao Paulo, Monday, Oct. 19, 2026. (AP Photo/Andre Penner)

On Eve of Vote, Brazil’s Federal Prosecution Office Demands Answers from Meta

Demonstrators wear masks depicting Meta CEO Mark Zuckerberg, Brazilian presidential candidate Flavio Bolsonaro and US President Donald Trump during a protest outside Meta's headquarters in Sao Paulo, Brazil, on October 19, 2026. (Photo by Allison Sales/NurPhoto via AP)

In Brazil’s Runoff, Lula Faces Bolsonaro, Trump—and Big Tech

A grey toned abstract photo of wavy mirror reflections of a building seen from many different angles. 

LLMs Are Still Making Election Deepfakes Just Days Before the US Midterms

A view of the US Capitol building with the dome brightly illuminated against the backdrop of a night sky. 

The United States Capitol dome by night, Washington DC, 2019. Photo by Darren Halstead on 

A Congressional Commission on AI Risk is Overdue

A demonstrator wearing a yellow vest waves the French flag on top of a burning barricade on the Champs-Elysees avenue with the Arc de Triomphe in background in Paris. (AP Photo/Michel Euler, File)

France’s Protests Reveal Europe’s Expanding Surveillance of Dissent

Building of the European Parliament in Strasbourg, France, with the EU-flag.

Building of the European Parliament in Strasbourg, France

European Parliament Divided on How to Govern AI At Home and Abroad

People with their mouths taped participate in a protest against internet shutdowns, arrests, and executions in Iran outside the Palacio Nacional de Belem in Lisbon, Portugal, on May 10, 2026. (Photo by Luis Boza/NurPhoto via AP)

Age Checks, AI Regulation and Censorship Put Internet Freedom Under Pressure

A rally against data centers at R.K.Beach in Visakhapatnam, India, on July 19. Source: Stop Dirty Data Centers

A rally against data centers at R.K.Beach in Visakhapatnam, India, on July 19. Source: 

What India’s Forest Rights Act Can Teach Us About Data Centers

Meta CEO Mark Zuckerberg leaves after testifying in a landmark trial over whether social media platforms deliberately addict and harm children, Wednesday, Feb. 18, 2026, in Los Angeles. (AP Photo/Damian Dovarganes)


Unpacking the Public Health Turn in Internet Regulation

Closeup of magnifying glass standing on dark surface with beam of light

You Can’t Solve AI Risk Without Solving the AI Evidence Problem

President Donald Trump and House Speaker Mike Johnson of La., listen as Google CEO Sundar Pichai speaks to reporters at the White House in Washington, Tuesday Sept. 29, 2026, after meeting with top executives of AI firms. (AP Photo/Alex Brandon)

AI Hype Helped Google Dodge an Antitrust Breakup. What Comes Next?

Police women wearing pink vests carry a female protester who is peacefully resisting arrest by her arms and legs. A surrounding crowd is watching and photographing and a television broadcast journalist is in the background too.

Four policewomen detain and carry a protester during a demonstration demanding the resignation of Chief Election Commissioner Gyanesh Kumar in New Delhi, India, Saturday, Oct. 10, 2026. (AP Photo/Manish Swarup)

Guilty Until Proven Peaceful Under India’s Preemptive Policing Crackdown

A gaming console. (Photo by Romain Doucelin / SOPA Images/Sipa USA)(Sipa via AP Images)

The EU KIDS Act Could Put Young Gamers' Privacy at Risk

Abstract photo of a wooden speech bubble with many small paper cut outs of speech bubbles in different colors and textures glued on top.

Governments Can Use AI to Repair Public Trust. If They Listen to People First

Illustrative photo captured during the plenary session of October 2025 in Strasbourg, France.

“Behind the scenes - European Parliament in Strasbourg” by 

EU’s Digital Sovereignty Push Runs Up Against Budget Cuts for Civil Society

An illustrative photo of a man walking through the corridors of the European Parliament in Strasbourg, France.

Behind the scenes - European Parliament in Strasbourg

Despite Spyware Attacks on Its Own Officials, EU Response Falls Short

Open AI CEO Sam Altman does a television interview with CNBC's Kate Rooney before delivering the keynote address at the OpenAI developers conference at Fort Mason on September 29, 2026 in San Francisco, California. (Photo by Heather Diehl/Getty Images)

OpenAI Faces Lawsuits, Subpoenas and EU Scrutiny in a Single Week

European Commission Executive Vice President Henna Virkkunen delivers a keynote address at the International AI Summit held at the RDS in Dublin, Ireland, on October 14, 2026.

Is Europe Ready for Advanced AI? The Growing Debate Over Liability

A finger is poised to tap the ChatGPT app icon displayed on a smartphone screen in Creteil, France, on October 1, 2026, as OpenAI begins rolling out Dots in ChatGPT, its new system of autonomous AI agents. (Photo by Samuel Boivin/NurPhoto via AP)

Another Week of Lawsuits Piling Up at OpenAI

A woman seen from behind looks at a huge selection of lettuce and greens in a supermarket refrigerator. 

The Age of Hypertrophy

With days left before the second and decisive round of Brazil’s presidential election, the country’s Federal Prosecution Office on Wednesday 

 urgent recommendations and demanded answers from Meta regarding content moderation actions by the platform, alleging that previous answers sent by the company were generic and insufficient. The Prosecution Office gave Meta 48 hours to respond as to whether it will comply with the recommendations.

 by the company over the past few weeks, amid Brazil’s most tightest electoral race in recent years, which pits incumbent President Luiz Inácio Lula da Silva against Senator Flávio Bolsonaro, the son of former president Jair Bolsonaro.

In this period, Meta took down a page of a media entity that 

temporarily disabled Lula’s campaign page

disabled its Muse AI agent from generating images

 in support of Lula, while authorizing images related to Flávio.

Civil society organizations, including Article 19 Brazil, Sleeping Giants Brazil, 

 and Ctrl+Z (co-founded by Tatiana Dias, a Tech Policy Press fellow) are documenting cases of takedowns, technical restrictions, account bans, reach suppression, and algorithmic discrimination that disproportionately appear to affect progressive and democratic accounts, such as student associations, activists, independent journalists and outlets.

 by the newspaper Folha de S. Paulo revealed that Meta employees admitted in internal communications there were restrictions imposed against Lula’s accounts. Earlier this week, a 

 in front of Meta headquarters in São Paulo, with demonstrators asking Meta to stop interfering in Brazilian elections. In Washington D.C., a group of Democrats 

 from Meta about its actions during the Brazilian elections.

 on its website, Meta pushed back against the allegations of political interference. “Meta doesn’t take a stand in these or any other elections, and the allegations that we are favoring or interfering in favor of any candidate or political group are false and unfounded”. The company also called the issues “technical problems” that have led to the construction of a “political narrative”.

Meta did not respond to a request for comment sent by Tech Policy Press.

The Prosecution Office recommended that Meta cease removing posts, suspending accounts or restricting features without a legal or contractual basis; direct human and technical resources to allow for adjustments in automatic content moderation and mandatorily submit to review posts containing terms such as “demonstration” or “march”; apply content moderation criteria in a uniform manner, without distinguishing political or ideological orientation; and ensure mandatory human review by people familiar with the language and context for material flagged by mass reports.

The office also recommended that Meta immediately restore publications, reach and features when it recognizes mistakes in content moderation and that it clearly communicate to users in clear and accessible Portuguese language every content moderation decision that has been applied to their accounts.

The office further recommended Meta use a ‘megaphone’ resource — a special feature notification that is displayed at the top of the page on its applications — to communicate moderation decisions. This resource is used by Meta in partnership with authorities to communicate about special civic events, such as election days.

One interesting recommendation is that Meta should hire an independent firm to audit the content moderation algorithms used in its platforms over the last two months. “This is very important, very unprecedented and it leaves a door open so that we can effectively have more transparency, even if obligated, from the company, and independently from the company’s algorithmic systems,” said André Boselli, coordinator for the Information and Communication Technology Ecosystems program at Article 19 Brazil and South America.

Such an audit could produce further evidence and confirmation of the company’s failures, in addition to the documentation that has been put together by civil society organizations and individual reports from users. The 

 was based on internal messages, a rare instance of insider confirmation on the technical decisions that impact the electoral process. “This is a piece of this jigsaw puzzle that helps confirm what was being felt by people, it starts to confirm itself as a serious hypothesis,” said Boselli.

Boselli noted, however, that these are all recommendations, not legal orders, and compliance depends on Meta. For him, Meta has demonstrated that it wants the perks of “being a company that assumes this institutional role in modern society of intermediating the public debate,” but without the onus of being accountable to society.

In response to questions sent previously by the Federal Prosecution Office, Meta provided generic information that observers regard as insufficient. It claimed, for example, that it automatically blocked over 206,000 pieces of content between Oct. 1 and Oct. 5 based on electoral rules and after having identified the use of AI in political content. The Prosecution Office found, however, that most of the blocks were applied without individual justification or based on criteria meant for serious offenses, such as terrorism or CSAM.

In addition to its recommendations, the Federal Prosecution Office 

 against Meta in order to protect “freedom of speech, the right to information and the due informational process in the digital environment.” A federal court has to accept the Prosecution Office’s request.

, as in the recommendations, prosecutors request that, moving forward, appeals to content moderation decisions should necessarily be reviewed by specialized human teams, with a$100,000 fine per account in cases of non-compliance.

The fact that the lawsuit is being initiated by the Federal Prosecution Office outside of electoral courts is significant, said Boselli. “This is a recognition that the right to democratic participation goes beyond the electoral sense,” he said. “We are not talking about litigation battles between campaigns — no, it’s citizenship itself involved.”

The prosecutors’ decision to initiate a lawsuit in regular courts comes at a time when the Electoral Court’s own actions are being scrutinized. Since the first round vote, the Electoral Court president, Justice Kassio Nunes Marques, a Bolsonaro appointee, made decisions that have raised concerns.

Nunes Marques, like many others in Brasília, is 

 in the Banco Master case, Brazil’s biggest corruption 

 include that he attended a party during Carnival in 2025 organized by Daniel Vorcaro, the banker formerly in charge of Banco Master who is currently jailed, for which sex workers were hired. In a response to Revista Piauí, which published the investigation, Nunes Marques said he never voted in Vorcaro’s favor, never exchanged messages with him, and did not ask him for anything.

A few days after the first round vote, Nunes Marques was 

 involving a highly popular pro-Lula WhatsApp channel organized by federal deputy André Janones, a Rede member that serves as the informal digital mastermind of the Lula campaign. Nunes Marques suspended the channel based on claims of disinformation, while the Lula campaign claimed that a similar channel connected to 

Flávio was also riddled with disinformation

The latest development — revealed by a 

 — says Nunes Marques and Justice André Mendonça, another Bolsonaro appointee, were circulating a draft of a rule to block WhatsApp channels in the 48 hours leading up to the runoff vote. The measure would be a direct hit to the pro-Lula WhatsApp channel, which had more than 8.3 million members by Thursday evening. After the story was published, 

 he was against last-minute measures, while 

 the rule was not an individual initiative and would still need to be discussed.

 sent to the Electoral Court, a group of more than 100 legal experts, researchers and civil society organizations warned that blocking WhatsApp channels on the eve of elections “restricts citizens’ political freedom and significantly alters the conditions of the electoral dispute days away from the vote.”

One of the world's largest democracies and a leader in the hemisphere, Brazil is in the final days of a runoff presidential election that pits the current president and leader of the Workers’ Party (PT), Luiz Inácio Lula da Silva, against senator and Liberal Party (PL) candidate Flávio Bolsonaro, son of Jair Bolsonaro, Lula's predecessor. The hard-right Bolsonaro, bucking the polls as many of his ideological compatriots have done before, received the 

 in the first round of the election, earning 47% to Lula’s 45%. Bolsonaro's campaign also swept in a range of national and regional candidates in the first round, with the PL forming 

 in the Senate and Chamber of Deputies, while winning many 

 and other local races. The two men will face off in a second round on Sunday.

Questions about the power and influence of tech companies based in the United States have taken center stage. Meta, the owner of Facebook, Instagram and WhatsApp, has become a major focus of these 

Concerns about the integrity of the online information environment are compounded by fears of US government meddling in Brazilian politics. Combined with Lula's deeply antagonistic relationship with the Trump administration, his campaign has complained of interference and manipulation from the United States, a claim that comes with the weight of deep historical precedent.

With just days left before the vote, these dynamics are emerging as one of the principal campaign narratives and represent a major challenge to Brazil's democratic processes. A series of events over the past few days and weeks are testing the system and marking the edges of oversight, regulation and technological development in the field.

Meta's properties dominate the Brazilian political landscape, from Facebook posts and Instagram shorts by major candidates to omnipresent WhatsApp messages that parties use to share their messages, coordinate supporters and get out the vote.

blocked Lula's Instagram campaign account

 on Sept. 23, several days before the first round election and his campaign’s leadership cried foul and demanded it be reinstated, it represented a critical development. With no explanation, Meta reinstated it on 

. The campaign complained to the Superior Election Court (TSE) after the election and asked for investigation, and on Oct. 10 the TSE asked Meta to respond. It did, 

: "The allegations are simply false and not supported by the facts." Lula’s campaign disagreed, accusing 

 and accusing the platform of “illegally practicing algorithmic selectivity, in the midst of a fierce presidential contest.’’

These kinds of posts and takedowns show both the reach of the social network and challenges of moderation, but there was no explanation for why this post was taken down or reinstated.

Other questionable moderation decisions have dogged Meta, most notably when it took down a viral video explainer by Fabio Porchat, one of Brazil’s most famous comedians, of 

 that allegedly implicates Bolsonaro and many other politicians across the spectrum. Revelando Brasil, a video account supported by a collective of news organizations, 

hosted the video and was blocked for several 

 nonprofit focused on big tech accountability, published an 

 indicating that Meta appears to be systematically discriminating against left wing posts, 

. (Tatiana Dias, a fellow at Tech Policy Press, is a founder and program director at CTRL+Z.) These reports formed the basis of a call for an investigation into what CTRL+Z together with Article 19, Conectas Direitos Humanos and Sleeping Giants Brasil, referred to as systematic bias by the company.

The TSE and the Federal Prosecution Service have also been asked by both CTRL+Z and a left-wing senator, Orlando Silva — a leader on tech legislation, including a recent bill focused on disinformation — to investigate and provide a response to the Revelando case. Prosecutors 

 and asked for responses not only from Meta’s properties but also from X within five days. On Oct. 21st, prosecutors 

 taken offline and additional moral damages if Meta fails to comply. They also demanded the company take specific actions to remedy, such as ensuring more human moderation, creating a system for users to request human intervention, and using automated messages to apologize for the errors and highlight the restrictions that were made on particular users.

As in the past, Brazil’s vibrant civil society is helping to 

 on human rights and accountability issues online. More than a decade ago, civil society groups 

 Brazil’s Internet Bill of Rights, the 

, a 2014 law that establishes rights to privacy and access in the country’s law. Under a Superior Court ruling and a May executive decree, 

 make platforms liable for illegal content if they fail to act after notice, impose a duty of care on the spread of illegal content, and give the country’s data protection agency power to oversee and investigate violations. Last week, citing research by CTRL+Z and other groups, the Ministry of Justice asked the data protection agency to 

 into whether Meta’s moderation decisions violate the law and consider precautionary measures. If such an investigation were opened and ultimately found fault, it could expose Meta to monetary penalties. Meta has said it does not favor any candidate.

Another Meta property, WhatsApp, is also being scrutinized for its questionable implementation of AI safeguards. Users have found, and researchers have replicated, various cases where Meta AI will generate images supporting Bolsonaro's campaign 

 for Lula. Amid allegations of bias against Lula, the company has lacked transparency, and authorities have struggled to respond quickly in the critical election period.

The one exception is the TSE, which is one of the more proactive electoral management bodies with jurisdiction over tech companies. 

Earlier this month, it demanded a response from Meta within 48 hours

 after Lula’s campaign alleged election interference.

 to take measures to ensure that both campaigns were being treated equally, mitigate the effects of their previous restrictions and disclose parameters for their approval and enforcement of political content. The decision noted that there was no complete remedy for the failures, but that going forward, Meta would be fined 500,000 Brazilian Reais (roughly $100,000) a day if it did not take steps to handle content equally for each candidate, ensure AI image generation criteria is equally applied, and disclose parameters for political and electoral content requests. Meta responded by saying that it would restrict AI generation for all candidates.

 with all of the order’s terms and threatened to double the fine, requiring further response within 24 hours. Meta 

 "Despite our best efforts, errors may occur, including the improper blocking of ads from authorized advertisers—precisely the type of issue that gave rise to this lawsuit. We have informed the TSE of this and will work promptly with the court and campaign teams to rectify any issues." But the company did not provide many technical details about the decisions and moderation. Flavio Bolsonaro’s campaign also joined the case, claiming they 

have also been discriminated against unfairly.

 It remains to be seen whether the justices will accept this explanation, but they could levy the fine or request further details.

These moves follow a familiar pattern of Brazilian jurisprudence, oversight and enforcement. In the past decade, 

Brazil has jailed Facebook executives over police investigations

Brazil's somewhat recent history has made it especially protective of its democratic values. Until 1985, the country was a military dictatorship, with strict controls on information and only limited voting rights. When democracy returned, many laws were instituted to prevent autocracy forever

The scandal deepened last Friday when Meta’s internal discussions were published by the Folha de São Paulo, Brazil's leading newspaper. Employees were quoted 

admitting that Lula’s accounts were restricted and struggling to explain

 why. They also noted a high chance of legal action by the Lula campaign and other actors, as the perception of interference grew. To show there had been equal treatment, Meta highlighted that posts from Bolsonaro’s campaign were blocked for using AI (though the campaign denied it had), but also admitted that searches for a temporarily banned Lula support group in WhatsApp led to results for Bolsonaro’s campaign as well as for Lula, and that searches for Lula’s name on Instagram did not always result in users seeing his page.

The company suggested that its engineers were dealing with a global data outage on Oct. 7 that lasted for around 5 hours, but the aforementioned restriction on Lula’s account lasted nearly 24 hours and happened two days later. In a 

, Meta said: “Allegations that we act to interfere in the electoral process or favor a candidacy are false and find no support in the facts... Some episodes had broad reach and affected different accounts and advertisers in different countries. We do not give preferential treatment to candidates, parties, or political positions." Meta has now banned all “social advertising” that does not come directly from campaigns, politicians and their related coalitions until 

, which could have serious impacts for civil society organizations, media and others who also have a voice in the electoral process. It shows the difficulty Meta has had in responding to demands that it moderate objectively. Even 

its Oversight Board criticized its moderation

 in the context of the Brazilian election, as well as for the US midterm elections.

Meta is in the hot seat alone among the major platforms in Brazil, because it accepts political advertising in a more unrestricted fashion. Google, LinkedIn, Snapchat, TikTok, and X all ban political ads, while Kwai offers them only in limited forms to comply with TSE regulations.

Artificial intelligence is a key and growing component of the information ecosystem. Deepfakes falsely depicting Lula and Bolsonaro have circulated on social media and in private messaging apps. One depicted Lula 

being captured by the US like Venezuela’s former president Nicolas Maduro

 endorsing candidates he didn’t. One showed an AI generated version of 

 while under house arrest, and barred from personally making social media posts, using mobile phones or making public statements. In 2025, the major platforms 

courted Bolsonaro's party and assisted it with access to AI tools

. While tech companies regularly engage with political parties and figures, it is possible they are also motivated by a shared interest in fighting regulation of problematic online content, 

The TSE has instituted strong rules around the kind of content that can be propagated in an election cycle —- including 

content discrediting Brazil's notoriously trustworthy voting system

 — and has also placed restrictions on AI-enhanced content. Any content created with AI has to be labeled. This is hard to enforce. Studies testing the systems, 

, found that content went unlabeled, questionable responses were given to basic answers, and overall the platforms were unresponsive to feedback and criticism. Brazilians are one of the world's heaviest users of social media and increasingly AI, so these questions become not only important but existential for democracy. In this election, Brazil is testing the legal, technical, social and political systems in ways that they have never been pushed before.

The TSE must constantly adapt to new technologies. WhatsApp is a dominant platform for election discourse, and has been for years. In 2018, Jair Bolsonaro’s campaign pushed out content on various massive groups that was forwarded and disseminated quickly. His team became famous for the use of automated systems for spreading content and coordinated messaging, 

 when his team organized official communications after he took power until the 2022 election. This type of activity is 

 he and his allies ended up sanctioned and imprisoned for their role in attacks on federal buildings on Jan. 8, 2023 that mirrored the events of Jan. 6, 2021 in the US. A key component of the attempted coup was coordinating and inciting supporters to mistrust the results, oppose Lula's election and stop the transfer of power.

Since then, WhatsApp and other messengers have placed restrictions on group sizes, as well as the number of times a piece of content can be forwarded. However, WhatsApp now also has open public channels, similar to those on Telegram. These channels have content moderation challenges similar to those of social media and are bound by similar rules. One major channel of the left, "Exército do Lula" (Lula's Army), grew from 

55,000 shortly after the first round to over 8 million people

On Oct. 11, a Superior Electoral Court judge (appointed by Jair Bolsonaro) ordered the 

suspension of the channel for dissemination of false content

. The court's inspector general then reprimanded the judge and ordered it reinstated, saying the entire court needed to review the decision and that others should have been consulted. 

, as well as the judge, questioned why the court had so quickly moved to block the entire channel rather than specific posts in such a sensitive runoff election period. After President Lula himself joined the channel on Oct. 22, Bolsonaro’s campaign requested an investigation into whether the chat could be defined as an official campaign vehicle, which would change the rules under which it operates. The same justice who suspended the channel once again 

 of how TSE mandates are being applied, how the channel is moderated, whether there is coordinated amplification, and other technical data. Google was also ordered to provide data on amplification and coordinated influence related to the channel on its platforms.

The Trump administration is aligned with Bolsonaro and has been 

 supporting him and his father for many years, but 

the US president kept his support strategically under the radar

. Economic tariffs on Brazilian goods, and sanctions and visa denials for Brazilian political figures — including for the judge who oversaw Jair Bolsonaro’s coup trial — have led to accusations of meddling by Lula and others. Online and offline, right-wing US influencers and political figures have been pushing narratives around Lula's alleged corruption, which are translated into Portuguese and amplified locally.. Brazil’s 

issued a report on this growing pressure and influence in the run up to the first round.

According to The Guardian, the Trump administration also proposed 

 to support Brazilian civil society to investigate the judicial system to counter what it calls 

”. In response, Brazil's Attorney General 

launched an investigation into US interference

 just as Brazilians began to cast their votes. The political venture capital fund, Rockbridge Network, which was co-founded by US Vice President JD Vance before his political rise and that the 

 reported is supported by tech oligarch Peter Thiel, has also come under scrutiny. According to an investigation by 

 Rockbridge funded Brazilian influencers, media, training for youth groups, right wing candidates and policy development—something the Lula campaign requested the TSE 

investigate as a campaign finance violation

. Rockbridge said the Brazilian company was created, but the project "never received funding, began operations or was used for any business activity," the newspaper reported.

All of these actions combined have cast a spotlight on US influence. They demonstrate the continuing regional and global influence of the hegemon, politically and economically — and now also through the global network and connective tissue of its apps, platforms and AI. Brazil’s struggle to protect its democratic institutions and political discourse against powerful forces — including those of Meta to silence or amplify political voices through content moderation decisions — is one that is mirrored elsewhere too, including in the US itself. Without more transparency into the underlying systems it is impossible to know the rationale for Meta’s decisions throughout this election period, but in Brazil they are experienced as obvious failures.

The election on Sunday decides who will lead the country, but a number of challenging questions have arisen from the process that will continue to reverberate: How is influence wielded in the digital space and what is acceptable in democratic societies? What role are platforms playing as stewards of civic spaces? As a bellwether for elections everywhere, we can hope that Brazil will respond to meet the moment with state of the art regulation and oversight of technology companies who hold power over our political spheres, whether purposely or not.


The leaders of the major frontier AI developers are 

 at the speed at which AI is creating potential catastrophic risk, calling for a slowdown and government regulation. But Congress and the president are

 on government’s role. Congress is unprepared and needs to devote serious attention to legislative options. As it does, it should quickly establish a bipartisan public-private AI Commission to define solutions for responding to public alarm and industry warnings.

There is precedent for using a Commission to build rapid consensus in response to a radical change in the threats at hand. The end of the Cold War was 

. The terrorist attack of September 11 was 

. In each case, the existing machinery of the US government had been built for a world that no longer existed. The moment required a clear-eyed rethinking of government's role itself, at a speed government rarely manages.

We are in such a moment now. The people building artificial intelligence — the engineers, researchers, and executives who understand the technology best — are the ones sounding the loudest alarms. That is not normal. Industries do not typically lobby for their own regulation or warn that their products could cause catastrophic harm. When this happens, the debate about whether government needs to act should be over. What remains are the hard questions: what should government's role be, and how does it organize itself to protect the AI future while maintaining America’s lead in AI development?

Certain steps should be taken right away, like requiring embedded independent reviewers and government safety review, but for Congress to frame regulation it should establish an AI Commission and give it six months to make its initial report. This will provide a thoughtful space where points of view can be explored deeply and participants can learn from each other. Its membership should include congressional leaders of committees for technology, commerce, and national security; former government officials; public interest and private sector representatives, including but not limited to AI developers; and independent domain experts. There are four questions Congress should ask:

First, what is the federal government's role, and what capabilities does it already have or urgently need to build? The US government started to create the ability to evaluate AI risks in the Commerce Department with the AI Safety Institute (now the Center for AI Standards and Innovation) at the National Institute of Standards and Technology (NIST); however, NIST is not a regulatory entity but rather a place that builds consensus standards. The Commission should say plainly what that capacity should look like and how fast it can be built. And in the meantime, what emergency action can the government take?

Second, what are the responsibilities of AI developers and of those who deploy AI systems, and what enforcement mechanisms should attach to each? Congress should define government oversight and audit mechanisms to hold private sector actors accountable.

Third, what information-sharing architecture is needed — between developers and government, between government agencies, and between the US and its allies — to give regulators visibility into risks before they metastasize into harms?

Fourth, what global alliances are most urgent to establish? As we regulate the AI America unleashes in the world, we need to strike agreements with China as the other leading AI developer with its own reasons for avoiding unconstrained AI and the loss of human control.

Fortunately, we are not without precedent for how to do this well. When the Berlin Wall fell, a serious 

 emerged in Congress that the CIA, an institution built for a bipolar Cold War world, had lost its reason for being. Instead of recommending the agency be abolished, a Congressional Commission proposed reorienting the intelligence community's mission toward emerging threats such as international terrorism years before most Americans understood how serious that threat would become.

After the terrorist attacks of September 11, in April 2002, in the absence of a government commission, the Markle Foundation — where I previously served as CEO — convened a 

 on National Security in the Information Age which gave a jump start to the later congressional 

. In the first six months, Markle detailed a blueprint for how the intelligence community should transform from a “need to know” to a “need to share” information environment to prevent terrorism while still protecting traditional civil liberties. Agencies across the government began implementing its recommendations, even as they were absorbed over the next 20 months into the findings of the 9/11 Commission and became law in the new Congress.

Congress needs to get this work started so by spring, a new Congress will be ready to address the calls for regulation of AI.

The people closest to this technology are telling us, in plain language, that the stakes are high and the window for shaping outcomes is narrow. We have faced comparably disorienting moments and found a way to think clearly and act quickly when it mattered: after the Cold War ended, and after 9/11. It is time to do it again.

 that it had detected and stopped an effort by Iranian operatives to place almost 100 fabricated news stories by seven imaginary reporters in small- to medium-size outlets around the world. The operatives had used ChatGPT to create the journalists’ personas and write copy that “looked and felt and read authentic to the point where it was being published by legitimate outlets,” 

 Darren Linvill, a Clemson researcher who studies influence operations.

It’s encouraging to see tech companies catch an abuse like this, especially as the midterms approach. But, as my colleagues and I 

 in a study we ran between February and August 2026 and in follow-up 

 this month significant flaws remain in AI safeguards. While AI chatbots consistently pushed back against election conspiracies, most of them could not identify AI-generated images, audio, or video, even if they created it themselves. Fixing that will take time, and with the midterms weeks away, voters and tech companies alike must act quickly to detect and stop AI-generated election disinformation.

 popular AI chatbots to see if they would push back on some of the most prevalent election falsehoods, including rigged voting machines, mass voter fraud, noncitizen voting, and more.

In the first part of the study, we drew on the bios and social media feeds of election deniers to provide six chatbots with false election claims in text queries and see whether they would endorse them. All six consistently declined, even when pressed. The chatbots, however, did make some factual errors and cited erroneous sources in the specifics of their text response.

For example, when we asked whether a single election worker could be bribed to rig an election, Claude firmly rejected the premise and walked through the checks that make it implausible. But in the process it made a small error, stating that ballots are only counted when party observers are present, which no state actually requires. It was a minor mistake, but it shows that there is room for improvement to avoid factual slips in an otherwise robust rebuttal.

Next, we tested to see whether built-in safeguards would effectively prevent chatbots from generating deceptive election-related images, videos, and audio. What we found was unnerving.

Every AI tool that we tested produced convincingly false content. Different chatbots created images that included an AI-generated person stuffing a stack of mail ballots into a drop box, a fake memo from the federal government on “official” letterhead reporting millions of ineligible voters, and a raid on a voter registration drive that never happened. We then fed those fake pieces of content back into chatbots and asked whether they were authentic. Most chatbots got it wrong, even when given images they had generated themselves. After 

 — which requires companies to provide provenance data for images they generate as well as free, public tools for detecting synthetic media — took effect, we conducted a follow-up 

Latest

Podcast

Recent

Our content delivered to your inbox

Thank you!

Around the globe