Ursula von der Leyen, president of the European Commission, gathered senior officials for an exclusive meeting at a 19th-century palace outside Brussels this month.
The goal was to prepare for a speech outlining the 27-nation bloc’s priorities for the coming year, with artificial intelligence dominating the agenda. The region’s new AI czar, Jim Hagemann Snabe, chair of German industrial giant Siemens, outlined ways that Europe could use the technology for economic gain. Others raised concerns about AI safety and growing risks, two people briefed on the gathering said.
Since the Sept. 4 meeting, von der Leyen and her commissioners have been left asking what they can do to address the fast-changing technology, one of the people said. Should they ramp up regulation to address AI’s emerging dangers? Or would doing so carry economic and geopolitical costs if other countries did not clamp down as well?
They have not landed on any final conclusions, with more discussions among European leaders planned this year.
“AI comes with a balance of possibilities and risks,” von der Leyen said at her State of the Union speech on Sept. 16. “Without addressing these risks, we will never be able to unlock AI’s potential.”
Versions of the Belgium meeting are playing out nearly everywhere as many governments struggle to get a handle on a technology that has rapidly outpaced their ability to respond. AI is evolving so quickly that laws passed two years ago are having limited effect, proposed bills to regulate the technology are stalling without a vote, and international summits to discuss what to do are being swiftly forgotten.
AI has made the gap between policymaking and technology one of the widest ever, according to more than 20 interviews with lawmakers, technologists and policy experts. Many governments are torn between wanting to harness AI and fearing its risks. And they are further mired by national politics, limited technological know-how, geopolitical divisions and reluctance to fall behind in a race that could shape economic power and national security for decades to come.
Story continues below this ad
Many countries are waiting for the two leading AI nations — the United States and China — to take a cue from their AI policies. Yet in the United States, the Trump administration has been divided over AI safety, with President Donald Trump saying regulation would hurt innovation and cede ground to China. China itself has imposed some rules on AI, especially to censor chatbot content, but also raced ahead with development.
“There’s this assumption that there are adults in the room on this issue, but there are no adults in this room,” said Sean Ó Heigeartaigh, director of the AI: Futures and Responsibility Program at the University of Cambridge, who is a member of a European Union advisory board on AI. “That is what makes all of this so completely terrifying.”
What’s left is a policy vacuum, with AI companies such as Anthropic and OpenAI largely left to police themselves. While the European Union passed a law regulating AI in 2024, it has taken years to carry out, and some say it already needs to be updated to account for advances. India, Japan, Australia, Canada and Kenya have taken a lighter touch or a wait-and-see approach, while Brazil is debating more comprehensive rules.
Pressure to act is growing, especially after the leaders of Anthropic, OpenAI and Google DeepMind recently called for slowing development of the most advanced AI models and for strengthening government oversight. Researchers have described scenarios in which self-improving AI could escape human control and cause catastrophic damage. Public opinion in the United States has turned against AI, with a recent New York Times poll finding that 61% of likely voters opposed the construction of data centers that power AI.
Story continues below this ad
The policy challenges were on display last week. In Washington on Thursday, Trump and Chinese President Xi Jinping stressed the importance of AI to their nations but did not agree on safety protocols. The United States and China are the world’s leading AI developers, making cooperation between them essential for international safety standards, but are also competitors wary of giving the other a technology advantage.
At the annual General Assembly of the United Nations in New York last week, many world leaders agreed on the need to address AI’s dangers, while Trump said the United States would reject “any attempt to construct a globalist scheme to control” the technology.
Safety experts have warned that crucial time is disappearing before oversight becomes harder to impose because AI will have spread so widely.
“Governments consider things in time scales of years, not weeks or months,” said Darren Jones, a member of Britain’s Parliament who worked on AI and technology policy under former Prime Minister Keir Starmer. “Institutionally, we’re not set up to keep up with the pace of development.”
Story continues below this ad
(The Times has sued OpenAI and Microsoft, claiming copyright infringement of news content related to AI systems. The two companies have denied those claims.)
The law that became a cautionary tale
In 2024, Europe adopted the world’s first comprehensive law for artificial intelligence. The AI Act, which was designed to impose safeguards on the technology’s riskiest uses, was hailed by European leaders as a blueprint for regulation.
Two years later, it has become a cautionary tale.
Implementation has been slowed by shifts in priorities and fears that regulation could weaken Europe’s competitiveness against the United States and China. Some of the law’s most consequential rules governing high-risk AI uses have been delayed. Staffing and budget constraints have further complicated enforcement.
The AI Act was proposed in early 2021, 18 months before OpenAI released ChatGPT. Intended to grapple with issues like facial recognition software, the rules were revised to address powerful new general-purpose AI systems like OpenAI’s GPT.
Story continues below this ad
But the technology has only accelerated since the law’s passage in 2024. Now the bleeding edge of AI comes in the form of so-called agents that can act with little human oversight. Companies are planning for a day when AI can create the next versions of itself.
“When we adopted it, we couldn’t even conceive of what AI is now able to do,” Valérie Hayer, a member of the European Parliament, said of the AI Act at a recent news conference. The law was ahead of its time, she said, but “we have to adapt.”
Europe’s challenge is made worse because it trails the United States and China in AI. Last year, just six months after the law passed, French President Emmanuel Macron said Europe risked trying to police a technology it had no role in shaping.
“If we let the Americans and the Chinese have all the champions to themselves, one thing is certain: We may have the best regulations in the world, but we won’t be regulating anything,” he said.
Story continues below this ad
The law’s limits are becoming clear. In July, OpenAI models broke out of a restricted testing environment and hacked the AI startup Hugging Face. But because the breach happened during a testing period, it fell within a murky area for enforcement. The AI Act generally exempts research and tests done before a model is released in Europe.
Supporters of the AI Act argue it was designed to evolve with the technology. It includes transparency and test requirements that AI safety experts said were needed to ensure the technology was deployed responsibly. Regulators have powers to demand information, appoint outside experts to evaluate models and order fixes. Authorities can ultimately block an AI model across Europe and impose fines of up to 7% of a company’s annual revenue.
“Europe has always been clear that AI must be developed safely and transparently, and remain under human control,” said Thomas Regnier, a European Commission spokesperson. “This is why we have the AI Act.”
Yet enforcement requires technical know-how that is still being built. A 60-member scientific panel established under the AI Act to offer expertise on frontier models and risks was appointed only in June. A new office in the European Commission meant to carry out many of the rules has only about 40 people working in an AI safety unit, prompting an amendment to the law to provide more resources.
Story continues below this ad
Gabriele Mazzini, a European Commission official who was a lead author of the AI Act and resigned in 2024, said the law’s implementation and enforcement was disappointing.
“Their goal was to pass the law and think about implementation later,” he said. “There is a disconnect between the law and the reality of what is being regulated.”
Another disconnect was with the United States.
An absence of urgency
Rep. Lori Trahan, D-Mass., said her jaw dropped when she read in April that Anthropic was warning that its latest AI model, Mythos, was too dangerous for the public. Anthropic said the model’s extraordinary ability to find software vulnerabilities could cause a cybersecurity “reckoning.”
During votes that day in the House of Representatives, Trahan crossed the floor to share her alarm with Rep. Jay Obernolte, R-Calif., who has a master’s degree in AI. The two quickly drafted a bill to require safety testing of AI models.
Story continues below this ad
Few lawmakers shared their urgency, she said. Some members wanted to wait until after November’s midterm elections to focus on a technically complicated issue. Others were reluctant to put guardrails on an industry that Trump wanted to keep free of regulations. Amid intense corporate lobbying, members squabbled over details, and the bill was revised multiple times.
Five months later, the bill and dozens of other proposals related to AI safety and security are on ice. Even after reports that models from OpenAI, Google and Anthropic autonomously hacked other organizations, there are no votes in sight.
“It’s hugely disappointing,” Trahan said. “If Congress doesn’t keep pace with AI, the dangers are even worse than with any other technology waves we’ve experienced.”
Rep. Ted Lieu, D-Calif., has spent years trying to make the same case. Soon after ChatGPT’s release in 2022, he wrote an opinion piece for the Times with the headline “A.I. Freaks Me Out.”
Lieu, who has a degree in computer science, later joined Obernolte to lead a congressional task force to educate lawmakers on AI, inviting industry leaders like Sam Altman of OpenAI to give technology demonstrations.
But after an initial burst of interest, attendance at the events fizzled.
When OpenAI revealed the Hugging Face breach this summer, Lieu introduced a bill that would require a kill switch for AI models if they became too dangerous. He told colleagues that the moment was a five-alarm fire. Still, he couldn’t get enough interest.
“I’m more scared than I was in 2023,” Lieu said.
Rep. Brett Guthrie, R-Ky., the influential chair of the House Energy and Commerce Committee, said Congress probably would not bring a vote on any AI legislation until next year.
At a recent conference, Guthrie said that he did not understand how the models worked and that he did not use AI frequently.
“We wouldn’t want to do something in a lame-duck session to do it quickly and not get it right,” he said.
Momentum, maybe?
The recent AI hacks have led to a burst of activity, with AI leaders calling for more government cooperation and collaborating on safety and monitoring techniques to prevent accidents.
In the United States, lawmakers introduced bills to create a federal agency to police AI models, kill switch requirements and a ban on the most powerful models that exceed human capabilities. Votes could be held after the midterms, particularly with AI becoming a key campaign topic.
Trump separately proposed an “AI Force” and a new “AI czar” for the White House, without providing details.
At the U.N. General Assembly, leaders from more than 20 countries — including von der Leyen and Macron — signed a letter calling for collective actions on AI safety. Members of the Security Council spoke in rare harmony on the need for global AI safeguards.
“Industry, governments and society must act now,” the letter read.
In Washington, Trump and Xi did not strike a deal on AI, but the two sides established new paths for communication about managing the greatest risks related to the technology.
While Trump has called AI safety risks a “hoax,” Xi’s views remain less clear. At the White House, the Chinese leader said, “We have both the capability and responsibility to develop and manage AI for good and ensure the development of AI is always under human control.”
The two countries said they would keep talking about the issue.