Anthropic is urging US states to pass stricter AI safety laws because the 2025 transparency rules are already too weak for current systems.
Cesar Fernandez, Anthropic’s head of US state and local government relations, told WIRED that self-reporting is no longer enough. “The transparency-focused safety bills of 2025 were a really important start, but as the capabilities of AI systems continue to advance quickly—the policy responses need to match,” he said. “We think that transparency and self reporting are no longer sufficient safety measures for the most powerful AI systems.”
It is unusual for a company valued at nearly $1 trillion to demand heavy rules. Anthropic’s founders believe the business must be built to ensure the world safely makes the transition through transformative AI.
Alongside California and New York laws, the company backed an Illinois measure requiring third-party auditors for AI labs. Most recently, it endorsed a Massachusetts policy with similar auditing rules, plus the power for the state attorney general to seek injunctive relief from non-compliant firms.
Fernandez joined Anthropic earlier this year. Before that, he led US state government relations for FanDuel and worked as a senior public policy associate at Uber. He helped both firms win policy battles in states across the country.
“Dubious” Motives
While safety groups and unions praise Anthropic’s stance, some Silicon Valley leaders see the political effort as a trap. David Sacks, the former White House AI czar and current technology adviser to President Donald Trump, claims Anthropic is using fear-mongering to secure its own position.
“Anthropic is running a sophisticated regulatory capture strategy based on fear-mongering,” Sacks wrote in a social media post last year. “It is principally responsible for the state regulatory frenzy that is damaging the startup ecosystem.”
Fernandez rejects the accusation. He notes that supported bills apply to “large AI model developers,” a term generally referring to companies with hundreds of millions in development spend and over $500 million in annual revenue. “It’s hard to imagine a startup meeting that threshold,” he says.
However, billions in capital are now needed for competitive frontier AI models. Safe Superintelligence, Thinking Machines Lab, and Mistral have each raised billions from investors, though their revenue remains far lower than Anthropic or OpenAI. These are potential competitors.
The more defensible version of the argument is that any company large enough to develop a powerful AI model should face the same rules because the underlying risks are identical. Fernandez notes that part of the goal is to “inspire a race to the top developing the most safe and secure AI systems,” which includes pushing for government legislation.
“Our mission is to make sure that this transition to a world with advanced AI goes well for Americans and humanity,” Fernandez says. “If you’re an AI model developer that’s developing powerful AI systems, there are requirements to be transparent with people on what you’re building, the model capabilities, and the risks associated with that model to critical infrastructure in states—all of that should be top priority.”
Where Anthropic Draws the Line
There are limits to what Anthropic thinks states should do without federal regulation. In a policy document published last month, the company recommended that governments should have a mechanism to block companies from deploying new AI models if they deem them unsafe.
The suggestion is ironic given the Trump administration recently told Anthropic to suspend access to its two most powerful AI models for foreign nationals, a move the company did not support. Anthropic believes the right to block unsafe deployments should be reserved for the federal government, though Fernandez notes it is an evolving conversation.
After the administration sent an export control directive that led to Mythos and Fable 5 models going offline, the company wrote in a blog post that blocking deployments should happen through a fair, transparent evaluation process.
Fernandez would not discuss the company’s federal policy work, but Anthropic has been active there. Last month, it sent a letter to the US government accusing Alibaba of a distillation attack—extracting information from Anthropic’s models through systematic prompting. Some researchers dismissed the claims as regulatory capture by another means.
They argue Anthropic’s real goal is to persuade the US government to ban open-weight Chinese models, which could prompt thousands of American businesses to turn to Anthropic’s enterprise offerings instead. Anthropic CEO Dario Amodei has notably warned Congress about the dangers of open source AI in the past.
At the state level, Fernandez says Anthropic has not specifically targeted open source AI models in endorsed legislation. “It’s less of a question of the model construction and more of the model capabilities, and when they rise to a certain level, they should be captured in these state frameworks,” he explains.
Whether you believe Anthropic’s efforts are sincere, the company is shaping the future of AI policy. Its latest model releases brought the cybersecurity capabilities of advanced AI into the national spotlight. Before that, the company had already spent years warning lawmakers about the catastrophic risks of advanced AI.
Concerns raised more often by American voters, such as losing jobs to the technology, the negative impacts of data centers in their communities, and the effects of chatbots on children, have not inspired a comparable legislative campaign from frontier AI labs. Anthropic and rivals have promised ordinary taxpayers won’t be stuck paying for data centers, and Anthropic has published proposals for responding to future AI-driven job displacement. The industry has put considerably less political muscle behind state laws addressing those problems.
Asked about that discrepancy, Fernandez says Anthropic is “eager to engage with lawmakers” on issues beyond catastrophic AI safety risks and is “having those conversations throughout the states.”
For now, however, those discussions have not produced the kind of coordinated state-level push Anthropic has mounted around the existential risks posed by frontier models.
What it means
Anthropic is betting that state laws are the only way to force safety standards before a federal system is in place. The company is willing to regulate its own competitors, but it draws a hard line at state-level deployment bans, reserving that power for the federal government. Meanwhile, the broader issues of data centre costs and job displacement are receiving far less attention from the industry.
