SAN FRANCISCO – Intelligence giant Anthropic is pursuing a one-civilization strategy that encourages nations to put in place stronger AI defenses, rather than sticking to a single set of rules.
That approach stands in stark contrast to the one favored by the OpenAI company, which has pushed state lawmakers to reach a common ground for regulating the breakthrough technology.
“While there are some in the industry who think of state policy as a way to create a higher level of federal law, Anthropic is not just looking to support the bill across the country in every state,” Cesar Fernandez, the firm’s head of U.S. government and local government relations, said in an interview with POLITICO on Tuesday. “We’re looking for legislation that raises the bar on security for more powerful AI systems.”
Fernandez’s comments came in response to questions from POLITICO about OpenAI’s ongoing campaign to create state AI regulations. ChatGPT’s chief influencer, Chris Lehane,he has introduced the word“change the federal system” to describe his attempts to bypass the paralyzed Congress and build a national AI system by mirroring state-by-state bills.
A veiled jab at OpenAI is at the Anthropic brand, whose executives left OpenAI in 2020 over concerns that the company was not prioritizing security. Anthropic has repeatedly pushed for strong AI safety laws at the federal and state levels — efforts that some critics,especially those close to the Trump administration and in venture capitalframed as an attempt to stretch controls and lock out competitors.
In a statement, OpenAI spokeswoman Liz Bourgeois defended the approach, saying it “changes the federal system, where effective government protections create national standards, help regulators enforce laws, provide the public with clearer protections, and allow developers to focus resources on security rather than conflicting requirements.”
The split between OpenAI and Anthropic’s approach to government houses comes at a critical time for AI regulation. With Congress reluctant to take action and the White Houseflipping between a light touch and a heavy handThe AI industry is increasingly looking to states for regulatory clarification. Whether state legislatures eventually come together on a single AI security framework or work to compete over time will have a big impact on the final shape of AI legislation in the US.
Like Lehane, Fernandez said he wants a federal system, but that the state’s response to the dangers posed by advanced AI models “cannot wait for action in Washington.”
The Anthropic lobbyist also set his company apart by touting its early intervention in government policy debates. Anthropic was the only leading AI labto approve California’s 2025 law to regulate advanced AI designs, the first such law in the country.
OpenAIhe did not take a standon California’s proposal before its passage. But it has since turned to legislation, which aims to promote greater transparency in companies’ security plans, as a model for other countries to emulate.
Anthropic, on the other hand, saw the California law as a springboard to revamp its efforts on AI safety. Fernandez said the rapid development of increasingly powerful AI models is the main reason his company has approved larger bills — in New York, Illinois and now Massachusetts — and its early testing in the legislative process.
“Each of those bills was stronger than the previous bill, and all of the bills advanced real safety obligations,” Fernandez said. “Transparency and self-reporting, we no longer believe are enough.”
He pointed to the powerful example of Anthropic’s Claude Mythos, which the company discovered was capable of exploiting security flaws in every major computer operating system during its tests. The cyber security concerns raised by Mythos (and its public-facing version, known as Fable) raised fears within the Trump administration, whichexport controls on technologyuntil Anthropic and the government can address the alleged vulnerabilities.
Late last year, OpenAI advocates successfully pressured New York Governor Kathy Hochul to amend her state’s AI safety bill to more closely match California law. But to the surprise of some safety advocates, it joined Anthropic in supporting the Illinois measure, which appears to be tougher than that of New York and California. The proposal, signed into law this month by Gov. JB Pritzker, requires leading AI companies to submit to an annual independent third-party review of their security plans — the first mandate of its kind.
Anthropic pushes the bar further. At the end of June, it isapproved regulationsunder development in Massachusetts for an economic development bond bill that Anthropic calls the nation’s strongest AI security proposal. The language it supported included a requirement for leading AI companies to hire independent evaluators to assess the potential for serious risks such as technology that aids in the development of biological weapons, as well as a provision giving the attorney general the authority to enforce the order.
Bourgeois, an OpenAI spokeswoman, said the company is still reviewing the Massachusetts proposal, but added OpenAI supports the state legislature’s view on AI protections.
AI bosses are also battled in the campaign. Each is associated withDuel networks super PACwhich has so far sunk tens of millions of dollars into political campaigns across the country. And in June, Anthropic began cutting checks directly to California lawmakers.
“We support candidates for election and re-election when their views on AI safety regulation are consistent with our commitment to ensure that the transition to a world with powerful AI works well for people in this country and around the world,” Fernandez said. “We strongly support candidates where there is ideological harmony.”
Fernandez said the company doesn’t cooperate with workers who eitherhave made donationsfor political candidates in California and elsewhere.
“We don’t direct our employees to make contributions, but they work at Anthropic because they care about the future of AI and where this is going if there is no proper security policy that has been enacted by the government,” said Fernandez. “I would think that drives them to get involved in the political process.”
