Anthropic escalates push for state AI regulations beyond California and New York

Anthropic is doubling down on its campaign for stricter state-level AI policies, expanding beyond the transparency laws it backed in California and New York last year to champion even more aggressive oversight measures. The company argues that existing frameworks focused on disclosure are now insufficient to manage the risks posed by increasingly powerful AI systems.

"We think that transparency and self reporting are no longer sufficient safety measures for the most powerful AI systems," said Cesar Fernandez, who leads Anthropic's government relations efforts, in recent remarks. The shift reflects the company's belief that regulatory guardrails must evolve as AI capabilities accelerate.

Growing Push for Stricter Rules#

Anthropic has recently thrown its weight behind proposals in multiple states that go beyond transparency requirements. The company endorsed an Illinois bill mandating independent third-party audits of AI safety practices, and backed a Massachusetts measure that would require similar external reviews while granting the state attorney general authority to seek court orders against non-compliant companies. This expansion signals Anthropic's conviction that frontier AI development demands more intrusive government oversight.

The strategy appears counterintuitive for a startup now valued near $1 trillion. Yet Anthropic's leadership has long viewed strict regulation as compatible with—perhaps even essential to—its core mission of ensuring safe AI development, a position that sets it apart from much of Silicon Valley.

Critics Cry "Regulatory Capture"#

Anthropic faces skepticism from rivals and Trump administration officials who suggest the company's lobbying efforts amount to competitive positioning masquerading as safety advocacy. David Sacks, the White House AI adviser, previously accused Anthropic of running a "regulatory capture strategy based on fear-mongering" designed to burden smaller competitors with compliance costs.

Fernandez rejects the charge, arguing that Anthropic only supports rules targeting "large AI model developers"—firms with hundreds of millions in development spending and over $500 million in annual revenue. He contends that startups are unlikely to clear such thresholds. Yet emerging competitors like Safe Superintelligence and Mistral have already raised billions, suggesting they could eventually face the same regulations, though their current revenues lag those of Anthropic and OpenAI.

A more credible framing of Fernandez's position, perhaps, is that any organization building powerful AI models should face identical safety requirements, since the underlying technological hazards remain constant regardless of company size. Anthropic frames its regulatory advocacy as spurring "a race to the top" in AI safety practices.