OpenAI’s CEO Sam Altman is stepping down from the internal committee that the company created to advise its board on “critical safety and security” decisions amid the race to develop ever more powerful artificial intelligence technology.

The committee, formed in May, had been evaluating OpenAI’s processes and safeguards over a 90-day period. OpenAI published the committee’s recommendations following the assessment on Sept. 16. First on the list: establishing independent governance for safety and security.

[time-brightcove not-tgx=”true”]

As such, Altman, who, in addition to serving OpenAI’s board, oversees the company’s business operations in his role as CEO, will no longer serve on the safety committee. In line with the committee’s recommendations, OpenAI says the newly independent committee will be chaired by Zico Kolter, Director of the Machine Learning Department at Carnegie Mellon University, who joined OpenAI’s board in August. Other members of the committee will include OpenAI board members Quora co-founder and CEO Adam D’Angelo, retired U.S. Army General Paul Nakasone, and former Sony Entertainment president Nicole Seligman. Along with Altman, OpenAI’s board chair Bret Taylor and several of the company’s technical and policy experts will also step down from the committee.

Read more: The TIME100 Most Influential People in AI 2024

The committee’s other recommendations include enhancing security measures, being transparent about OpenAI’s work, and unifying the company’s safety frameworks. It also said it would explore more opportunities to collaborate with external organizations, like those used to evaluate OpenAI’s recently released series of reasoning models o1 for dangerous capabilities.

The Safety and Security Committee is not OpenAI’s first stab at creating independent oversight. OpenAI’s for-profit arm, created in 2019, is controlled by a non-profit entity with a “majority independent” board, tasked with ensuring it acts in accordance with its mission of developing safe broadly beneficial artificial general intelligence (AGI)—a system that surpasses humans in most regards.

In November, OpenAI’s board fired Altman, saying that he had not been “consistently candid in his communications with the board, hindering its ability to exercise its responsibilities.” After employees and investors revolted—and board member and company president Greg Brockman resigned—he was swiftly reinstated as CEO, and board members Helen Toner, Tasha McCauley, and Ilya Sutskever resigned. Brockman later returned as president of the company.

Read more: A Timeline of All the Recent Accusations Leveled at OpenAI and Sam Altman

The incident highlighted a key challenge for the rapidly growing company. Critics including Toner and McCauley argue that having a formally independent board isn’t enough of a counterbalance to the strong profit incentives the company faces. Earlier this month, Reuters reported that OpenAI’s ongoing fundraising efforts, which could catapult its valuation to $150 billion, might hinge on changing its corporate structure.

Toner and McCauley say board independence doesn’t go far enough and that governments must play an active role in regulating AI. “Even with the best of intentions, without external oversight, this kind of self-regulation will end up unenforceable,” the former board members wrote in the Economist in May, reflecting on OpenAI’s November boardroom debacle. 

In the past, Altman has urged regulation of AI systems, but OpenAI also lobbied against California’s AI bill, which would mandate safety protocols for developers. Going against the company’s position, more than 30 current and former OpenAI employees have publicly supported the bill.

The Safety and Security Committee’s establishment in late May followed a particularly tumultuous month for OpenAI. Ilya Sutskever and Jan Leike, the two leaders of the company’s “superalignment” team, which focused on ensuring that if AI systems surpass human-level intelligence, they remain under human control, resigned. Leike accused OpenAI of prioritizing “shiny products” over safety in a post on X. The team was disbanded following their departure. The same month, OpenAI came under fire for asking departing employees to sign agreements that prevented them from criticizing the company or forfeit their vested equity. (OpenAI later said that these provisions had not and would not be enforced and that they would be removed from all exit paperwork going forward).

On August 7, a group of renowned professors co-authored a letter urging key lawmakers to support a California AI bill as it enters the final stages of the state’s legislative process. In a letter shared exclusively with TIME, Yoshua Bengio, Geoffrey Hinton, Lawrence Lessig, and Stuart Russell argue that the next generation of AI systems pose “severe risks” if “developed without sufficient care and oversight,” and describe the bill as the “bare minimum for effective regulation of this technology.”

[time-brightcove not-tgx=”true”]

The bill, titled the Safe and Secure Innovation for Frontier Artificial Intelligence Models Act, was introduced by Senator Scott Wiener in February of this year. It requires AI companies training large-scale models to conduct rigorous safety testing for potentially dangerous capabilities and implement comprehensive safety measures to mitigate risks.

“There are fewer regulations on AI systems that could pose catastrophic risks than on sandwich shops or hairdressers,“ the four experts write.

The letter is addressed to the respective leaders of the legislative bodies the bill must pass through if it is to become law: Mike McGuire, the president pro tempore of California’s senate, where the bill passed in May; Robert Rivas, speaker of the state assembly, where the bill will face a vote later this month; and state Governor Gavin Newsom, who—if the bill passes in the assembly—must sign or veto the proposed legislation by the end of September.

With Congress gridlocked and Republicans pledging to reverse Biden’s AI executive order if elected in November, California—the world’s fifth-largest economy and home to many of the world’s leading AI developers—plays what the authors see as an “indispensable role” in regulating AI. If passed, the bill would apply to all companies operating in the state.

While polls suggest the bill is supported by the majority of Californians, it has been subject to harsh opposition from industry groups and tech investors, who claim it would stifle innovation, harm the open-source community, and “let China take the lead on AI development.” Venture capital firm Andreessen Horowitz has been particularly critical of the bill, setting up a website that urges citizens to write to the legislature in opposition. Others, such as startup incubator YCombinator, Meta’s Chief AI Scientist Yann LeCun, and Stanford professor Fei-Fei Li (whose new $1 billion startup has received funding from Andreessen Horowitz) have also been vocal in their opposition.

The pushback has centered around provisions in the bill which would compel developers to provide reasonable assurances that an AI model will not pose unreasonable risk of causing “critical harms,” such as aiding in the creation of weapons of mass destruction or causing severe damage to critical infrastructure. The bill would only apply to systems that both cost over $100 million dollars to train and are trained using an amount of computing power above a specified threshold. These dual requirements imply the bill would likely only affect the largest AI developers. “No currently existing system would be classified,” Lennart Heim, a researcher at the RAND Corporation’s Technology and Security Policy Center, told TIME in June.

“As some of the experts who understand these systems most, we can say confidently that these risks are probable and significant enough to make safety testing and common-sense precautions necessary,” the authors of the letter write. Bengio and Hinton, who have previously supported the bill, are both winners of the Turing Award, and often referred to as “godfathers of AI,” alongside Yann LeCun. Russell has written a textbook—Artificial Intelligence: A Modern Approach—that is widely considered to be the standard textbook on AI. And Lessig, a Professor of Law at Harvard, is broadly regarded as a founding figure of Internet law and a pioneer in the free culture movement, having founded the Creative Commons and authored influential books on copyright and technology law. In addition to the risks noted above, they cite risks posed by autonomous AI agents that could act without human oversight among their concerns.

Read More: Yoshua Bengio Is on the 2024 TIME100 List

“I worry that technology companies will not solve these significant risks on their own while locked in their race for market share and profit maximization. That’s why we need some rules for those who are at the frontier of this race,” Bengio told TIME over email.  

The letter rejects the notion that the bill would hamper innovation, stating that as written, the bill only applies to the largest AI models; that large AI developers have already made voluntary commitments to undertake many of the safety measures outlined in the bill; and that similar regulations in Europe and China are in fact more restrictive than SB 1047. It also praises the bill for its “robust whistleblower protections” for AI lab employees who report safety concerns, which are increasingly seen as necessary given reports of reckless behavior on the part of some labs.

In an interview with Vox last month, Senator Wiener noted that the bill has already been amended in response to criticism from the open-source community. The current version exempts original developers from shutdown requirements once a model is no longer in their control, and limits their liability when others make significant modifications to their models, effectively treating significantly modified versions as new models. Despite this, some critics believe the bill would require open-source models to have a “kill switch.”

“Relative to the scale of risks we are facing, this is a remarkably light-touch piece of legislation,” the letter says, noting that the bill does not have a licensing regime or require companies to receive permission from a government agency before training a model, and relies on self-assessments of risk. The authors further write: “It would be a historic mistake to strike out the basic measures of this bill.” 

Over email, Lessig adds “Governor Newsom will have the opportunity to cement California as a national first-mover in regulating AI. Legislation in California would meet an urgent need. With a critical mass of the top AI firms based in California, there is no better place to take an early lead on regulating this emerging technology.” 

When the U.K. hosted the world’s first AI Safety Summit last November, Rishi Sunak, the then Prime Minister, said the achievements at the event would “tip the balance in favor of humanity.” At the two-day event, held in the cradle of modern computing, Bletchley Park, AI labs committed to share their models with governments before public release, and 29 countries pledged to collaborate on mitigating risks from artificial intelligence. It was part of the Sunak-led Conservative government’s effort to position the U.K. as a leader in artificial intelligence governance, which also involved establishing the world’s first AI Safety Institute—a government body tasked with evaluating models for potentially dangerous capabilities. While the U.S. and other allied nations subsequently set up their own similar institutes, the U.K. institute boasts 10 times the funding of its American counterpart. 

[time-brightcove not-tgx=”true”]

Eight months later, on July 5, after a landslide loss to the Labour Party, Sunak left office and the newly elected Prime Minister Keir Starmer began forming his new government. His approach to AI has been described as potentially tougher than Sunak’s.  

Starmer appointed Peter Kyle as science and technology minister, giving the lawmaker oversight of the U.K.’s AI policy at a crucial moment, as governments around the world grapple with how to foster innovation and regulate the rapidly developing technology. Following the election result, Kyle told the BBC that “unlocking the benefits of artificial intelligence is personal,” saying the advanced medical scans now being developed could have helped detect his late mother’s lung cancer before it became fatal.

Alongside the potential benefits of AI, the Labour government will need to balance concerns from the public. An August poll of over 4,000 members of the British public conducted by the Centre for Data Ethics and Innovation found 45% respondents believed AI taking people’s jobs represented one of the biggest risks posed by the technology; 34% believed loss in human creativity and problem solving was one of the greatest risks.

Here’s what we know so far about Labour’s approach to artificial intelligence.

Regulating AI

One of the key issues for the Labour government to tackle will likely be how to regulate AI companies and AI-generated content. Under the previous Conservative-led administration, the Department for Science, Innovation and Technology (DSIT) held off on implementing rules, saying that “introducing binding measures too soon, even if highly targeted, could fail to effectively address risks, quickly become out of date, or stifle innovation and prevent people from across the UK from benefiting from AI,” in a 2024 policy paper about AI regulation. Labour has signaled a different approach, promising in its manifesto to introduce “binding regulation on the handful of companies developing the most powerful AI models,” suggesting a greater willingness to intervene in the rapidly evolving technology’s development.

Read More: U.S., U.K. Announce Partnership to Safety Test AI Models

Labour has also pledged to ban sexually explicit deepfakes. Unlike proposed legislation in the U.S., which would allow victims to sue those who create non-consensual deepfakes, Labour has considered a proposal by Labour Together, a think-tank with close ties to the current Labour Party, to impose restrictions on developers by outlawing so-called nudification tools. 

While AI developers have made agreements to share information with the AI Safety Institute on a voluntary basis, Kyle said in a February interview with the BBC that Labour would make that information-sharing agreement a “statutory code.”

Read More: To Stop AI Killing Us All, First Regulate Deepfakes, Says Researcher Connor Leahy

“We would compel by law, those test data results to be released to the government,” Kyle said in the interview.

Timing regulation is a careful balancing act, says Sandra Wachter, a professor of technology and regulation at the Oxford Internet Institute.

“The art form is to be right on time with law. That means not too early, not too late,” she says. “The last thing that you want is a hastily thrown together policy that stifles innovation and does not protect human rights.”

Watchter says that striking the right balance on regulation will require the government to be in “constant conversation” with stakeholders such as those within the tech industry to ensure the government has an inside view of what is happening at the cutting edge of AI development when formulating policy. 

Kirsty Innes, director of technology policy at Labour Together points to the U.K. Online Safety Act, which was signed into law last October as a cautionary tale of regulation failing to keep pace with technology. The law, which aims to protect children from harmful content online, took 6 years from the initial proposal being made to finally being signed in.

“During [those 6 years] people’s experiences online transformed radically. It doesn’t make sense for that to be your main way of responding to changes in society brought by technology,” she says. “You’ve got to be much quicker about it now.”

Read More: The 3 Most Important AI Policy Milestones of 2023

There may be lessons for the U.K. to learn from the E.U. AI Act, Europe’s comprehensive regulatory framework passed in March, which will come into force on August 1 and become fully applicable to AI developers in 2026. Innes says that mimicking the E.U. is not Labour’s endgame. The European law outlines a tiered risk classification for AI use cases, banning systems deemed to pose unacceptable risks, such as social scoring systems, while placing obligations on providers of high-risk applications like those used for critical infrastructure. Systems said to pose limited or minimal risk face fewer requirements. Additionally, it sets out rules for “general-purpose AI”, which are systems with a wide range of uses, like those underpinning chatbots such as OpenAI’s ChatGPT. General-purpose systems trained on large amounts of computing power—such as GPT-4—are said to pose “systemic risk,” and developers will be required to perform risk assessments as well as track and report serious incidents.

“I think there is an opportunity for the U.K. to tread a nuanced middle ground somewhere between a very hands-off U.S. approach and a very regulatory heavy E.U. approach,” says Innes.

Read More: There’s an AI Lobbying Frenzy in Washington. Big Tech Is Dominating

In a bid to occupy that middle ground, Labour has pledged to create what it calls the Regulatory Innovation Office, a new government body that will aim to accelerate regulatory decisions.

“Part of the idea of the Regulatory Innovation Office is to help regulators develop the capacity that they need a bit quicker and to give them the kind of stimulus and the nudge to be more agile,” says Innes.

A ‘pro-innovation’ approach

In addition to helping the government respond more quickly to the fast-moving technology, Labour says the “pro-innovation” regulatory body will speed up approvals to help new technologies get licensed faster. The party said in its manifesto that it would implement AI into healthcare to “transform the speed and accuracy of diagnostic services, saving potentially thousands of lives.”

Healthcare is just one area where Kyle hopes to use AI. On July 8, he announced the revamp of the DSIT, which will bring on AI experts to explore ways to improve public services.

Meanwhile former Labour Prime Minister Tony Blair has encouraged the new government to embrace AI to improve the country’s welfare system. A July 9 report by his think tank the Tony Blair Institute for Global Change, concluded AI could save the U.K. Department for Work and Pensions more than $1 billion annually.

Blair has emphasized AI’s importance. “Leave aside the geopolitics, and war, and America and China, and all the rest of it. This revolution is going to change everything about our society, our economy, the way we live, the way we interact with each other,” Blair said, speaking on the Dwarkesh Podcast in June.

Read more: How a New U.N. Advisory Group Wants to Transform AI Governance

Modernizing public services is part of Labour’s wider strategy to leverage AI to grow the U.K. tech sector. Other measures include making it easier to set up data centers in the U.K., creating a national data library to bring existing research programs together, and offering decade-long research and development funding cycles to support universities and start-ups.

Speaking to business and tech leaders in London last March, Kyle said he wanted to support “the next 10 DeepMinds to start up and scale up here within the U.K.” 

Workers’ rights

Artificial intelligence-powered tools can be used to monitor worker performance, such as grading call center-employees on how closely they stick to the script. Labour has committed to ensuring that new surveillance technologies won’t find their way into the workplace without consultation with workers. The party has also promised to “protect good jobs” but, beyond committing to engage with workers, has offered few details on how. 

Read More: As Employers Embrace AI, Workers Fret—and Seek Input

“That might sound broad brush, but actually a big failure of the last government’s approach was that the voice of the workforce was excluded from discussions,” says Nicola Smith, head of rights at the Trades Union Congress, a union-group.  

While Starmer’s new government has a number of urgent matters to prioritize, from setting out its legislative plan for year one to dealing with overcrowded prisons, the way it handles AI could have far-reaching implications.

“I’m constantly saying to my own party, the Labour Party [that] ‘you’ve got to focus on this technology revolution. It’s not an afterthought,” Blair said on the Dwarkesh Podcast in June. “It’s the single biggest thing that’s happening in the world today.”

On June 8, Republicans adopted a new party platform ahead of a possible second term for former President Donald Trump. Buried among the updated policy positions on abortion, immigration, and crime, the document contains a provision that has some artificial intelligence experts worried: it vows to scrap President Joe Biden’s executive order on AI.

[time-brightcove not-tgx=”true”]

“We will repeal Joe Biden’s dangerous Executive Order that hinders AI Innovation, and imposes Radical Leftwing ideas on the development of this technology,” the platform reads.

Biden’s executive order on AI, signed last October, sought to tackle threats the new technology could pose to civil rights, privacy, and national security, while promoting innovation and competition and the use of AI for public services. It requires developers of the most powerful AI systems to share their safety test results with the U.S. government and calls on federal agencies to develop guidelines for the responsible use of AI domains such as criminal justice and federal benefits programs.

Read More: Why Biden’s AI Executive Order Only Goes So Far

Carl Szabo, vice president of industry group NetChoice, which counts Google, Meta, and Amazon among its members, welcomes the possibility of the executive order’s repeal, saying, “It would be good for Americans and innovators.”

“Rather than enforcing existing rules that can be applied to AI tech, Biden’s Executive Order merely forces bureaucrats to create new, complex burdens on small businesses and innovators trying to enter the marketplace. Over-regulating like this risks derailing AI’s incredible potential for progress and ceding America’s technological edge to competitors like China,” said Szabo in a statement.

However, recent polling shared exclusively with TIME indicates that Americans on both sides of the political aisle are skeptical that the U.S. should avoid regulating AI in an effort to outcompete China. According to the poll conducted in late June by the AI Policy Institute (AIPI), 75% of Democrats and 75% of Republicans believe that “taking a careful controlled approach” to AI is preferable to “moving forward on AI as fast as possible to be the first country to get extremely powerful AI.”

Dan Hendrycks, director of the Center for Safe AI, says, “AI safety and risks to national security are bipartisan issues. Poll after poll shows Democrats and Republicans want AI safety legislation.”

Read more: U.S. Voters Value Safe AI Development Over Racing Against China, Poll Shows

The proposal to remove the guardrails put in place by Biden’s executive order runs counter to the public’s broad support for a measured approach to AI, and it has prompted concern among experts. Amba Kak, co-executive director of the AI Now Institute and former senior advisor on AI at the Federal Trade Commission, says Biden’s order was “one of the biggest achievements in the last decade in AI policy,” and that scrapping the order would “feel like going back to ground zero.” Kak says that Trump’s pledge to support AI development rooted in “human flourishing” is a subtle but pernicious departure from more established frameworks like human rights and civil liberties.

Ami Fields-Meyer, a former White House senior policy advisor on AI who worked on Biden’s executive order, says, “I think the Trump message on AI is, ‘You’re on your own,’” referring to how repealing the executive order would end provisions aimed at protecting people from bias or unfair decision-making from AI.

NetChoice and a number of think tanks and tech lobbyists have railed against the executive order since its introduction, arguing it could stifle innovation. In December, venture capitalist and prominent AI investor Ben Horowitz criticized efforts to regulate “math, FLOPs and R&D,” alluding to the compute thresholds set by Biden’s executive order. Horowitz said his firm would “support like-minded candidates and oppose candidates who aim to kill America’s advanced technological future.”

While Trump has previously accused tech companies like Google, Amazon, and Twitter of working against him, in June, speaking on Logan Paul’s podcast, Trump said that the “tech guys” in California gave him $12 million for his campaign. “They gave me a lot of money. They’ve never been into doing that,” Trump said.

The Trump campaign did not respond to a request for comment.

Even if Trump is re-elected and does repeal Biden’s executive order, some changes wouldn’t be felt right away. Most of the leading AI companies agreed to voluntarily share safety testing information with governments at an international summit on AI in Seoul last May, meaning that removing the requirements to share information under the executive order may not have an immediate effect on national security. But Fields-Meyer says, “If the Trump campaign believes that the rigorous national security safeguards proposed in the executive order are radical liberal ideas, that should be concerning to every American.”

Fields-Meyer says the back and forth over the executive order underscores the importance of passing federal legislation on AI, which “would bring a lot more stability to AI policy.” There are currently over 80 bills relating to AI in Congress, but it seems unlikely any of them will become law in the near future.

Sandra Wachter, a professor of technology regulation at the Oxford Internet Institute says Biden’s executive order was “a seminal step towards ensuring ethical AI and is very much on par with global developments in the UK, the EU, Canada, South Korea, Japan, Singapore and the rest of the world.” She says she worries it will be repealed before it has had a chance to have a lasting impact. “It would be a very big loss and a big missed opportunity if the framework was to be scrapped and AI governance to be reduced to a partisan issue,” she says. “This is not a political problem, this is a human problem—and a global one at that.”

Correction, July 11

The original version of this story misidentified a group that has spoken out against Biden’s executive order. It is NetChoice, not TechNet.

Brazil’s national data protection authority has ordered Meta to halt the use of data originating from the country to train its AI models.

Meta’s current privacy policy enables the company to use data from its platforms, including Facebook, Instagram, and WhatsApp to train its artificial intelligence models. However, that practice will no longer be permitted in Brazil after its national data protection authority gave the company five days to change its policy on Tuesday.

[time-brightcove not-tgx=”true”]

Brazil said the company will need to confirm it has stopped using the data or face a daily non-compliance fine of $50,000 Brazilian Reals (almost $9000), citing “the imminent risk of serious and irreparable or difficult-to-repair damage to the fundamental rights of the affected data subjects.”

Meta said it was “disappointed” with the Brazilian authority’s decision, saying it was a “step backward for innovation.”

“AI training is not unique to our services, and we’re more transparent than many of our industry counterparts who have been using public content to train their models and products,” the company tells TIME Wednesday, following the Brazilian authority’s decision.

The decision follows a report published in June by Human Rights Watch, which found that a popular dataset of images scrapped from online sources used to train image models, made by German nonprofit LAION, contained identifiable images of Brazilian children, which the report says places them at risk of deep fakes or other forms of exploitation. Human Rights Watch says they found 170 photos of children from at least 10 Brazilian states by reviewing less than 0.0001 percent of the images in the dataset.

Brazil is one of Meta’s biggest markets, with over 112 million Facebook users alone. In June at a conference in the South American country, Meta unveiled new AI tools for businesses on its WhatsApp platform.

The Brazilian authority said users were not sufficiently warned about the changes, and that the process for opting out was “not very intuitive.” Meta says their approach complies with local privacy laws, and that it will continue to address the Brazilian authority’s questions.

Brazil’s decision to stop Meta feeding user’s data into its AI models follows similar pushback in Europe. Last month, Meta delayed the launch of its AI services and paused plans to train its models on EU and U.K. data after receiving a complaint from the Irish privacy regulator. Meta is expected to push ahead with training in the U.S., which lacks federal online privacy protections.

Read more: Meta Faces Norwegian Complaint Over Plans to Train AI on User Images and Posts

This is not the first time Meta has found itself at odds with Brazilian authorities. In February, the company was barred from using its name in Brazil due to confusion with another company. Meta successfully overturned the decision in March.