Anthropic threw its help behind the primary wave of frontier AI safety laws in the US final yr, securing new transparency necessities in California and New York that a lot of Silicon Valley fought in opposition to, arguing they’d stifle the AI boom. However Anthropic says these legal guidelines could already be outdated, and the corporate is now pushing states to undertake even more durable rules.
“The transparency-focused security payments of 2025 had been a extremely vital begin, however because the capabilities of AI methods proceed to advance shortly—the coverage responses have to match,” Cesar Fernandez, Anthropic’s head of US state and native authorities relations, advised WIRED in an interview. “We expect that transparency and self reporting are now not ample security measures for essentially the most highly effective AI methods.”
Being so pro-regulation is an odd message to return from a startup that’s now valued at nearly $1 trillion. However Anthropic is an odd firm. As we’ve written before, Anthropic’s leaders consider it must construct an enormous enterprise primarily based on growing and promoting entry to superior synthetic intelligence to satisfy its founding mission: “to make sure that the world safely makes the transition by means of transformative AI.”
As Anthropic has grown, it’s began endorsing a few of the nation’s harshest proposed rules on frontier AI corporations. Many of those guidelines are designed to mitigate catastrophic AI dangers, together with the likelihood that superior fashions might contribute to monetary disasters or mass deaths.
Past the self-reporting legal guidelines in California and New York, Anthropic has additionally supported an Illinois measure requiring AI labs to get their security processes evaluated by third-party auditors. Most just lately, the corporate endorsed a Massachusetts policy that might additionally require third-party auditing for AI labs—and empower the state’s lawyer common to hunt injunctive aid from corporations that do not comply.
I sat down with Fernandez this week to know the place the corporate’s AI insurance policies are heading. Fernandez joined Anthropic earlier this yr, after beforehand serving as head of US state authorities relations for the sports activities betting big FanDuel and as a senior public coverage affiliate at Uber. He helped each corporations win coverage battles in states across the nation. His experience will probably show invaluable to Anthropic, as Congress continues to stall on passing AI regulation and states are taking the lead.
“Doubtful” Motives
Whereas Anthropic’s pro-regulation agenda has been praised by AI security teams, labor unions, and different allies of the corporate, some Silicon Valley leaders interpret its political efforts by means of a extra nefarious lens.
David Sacks, the previous White Home AI czar and present know-how adviser to President Donald Trump, has claimed that Anthropic is basically making an attempt to get cumbersome legal guidelines handed to lure smaller AI startups in purple tape, thus securing its personal place as a frontrunner within the AI race.
“Anthropic is working a classy regulatory seize technique primarily based on fear-mongering,” Sacks wrote in a social media post final yr. “It’s principally chargeable for the state regulatory frenzy that’s damaging the startup ecosystem.”
Fernandez outrightly denies the accusation, noting that Anthropic has solely supported state AI payments that apply to “giant AI mannequin builders,” a time period outlined in a different way in each invoice, however that usually refers to corporations which have spent tons of of thousands and thousands of {dollars} on AI improvement and have greater than $500 million in annual income. “It’s laborious to think about a startup assembly that threshold,” Fernandez says.
However in a world the place billions of {dollars} in capital is now required to develop a aggressive frontier AI mannequin, there’s arguably a handful of promising startups that might meet these thresholds quickly sufficient. To call a number of, Safe Superintelligence, Thinking Machines Lab, and Mistral have every raised billions of {dollars} from traders, although their income stays far decrease than the likes of Anthropic or OpenAI. After all, these aren’t your run-of-the-mill startups, however they’re potential opponents of Anthropic.
