Anthropic Pushes States for Tougher AI Safety Laws
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Anthropic is lobbying states to go beyond the 2025 transparency-focused safety bills it helped pass in California and New York, with state policy lead Cesar Fernandez telling WIRED that transparency and self-reporting are "no longer sufficient" for the most powerful AI systems.
- Anthropic has backed an Illinois bill requiring third-party safety audits of AI labs and a Massachusetts measure that would also mandate third-party auditing and let the state attorney general seek injunctive relief against noncompliant companies.
- David Sacks, President Trump's AI adviser and former White House AI czar, accused Anthropic of running a "sophisticated regulatory capture strategy" aimed at bogging down smaller startups; Fernandez counters that the bills target only companies with roughly $500 million or more in annual revenue and hundreds of millions spent on AI development.
- Anthropic published a policy document recommending the authority to block unsafe model deployments be reserved for the federal government, not states — a stance drawing attention after the Trump administration recently ordered the company to suspend access to its Mythos and Fable 5 models for foreign nationals.
- Anthropic recently accused Chinese tech giant Alibaba of a "distillation attack" by systematically extracting information from its models, though some AI researchers have dismissed the claims as regulatory capture aimed at banning open-weight Chinese models that compete with Anthropic's enterprise offerings.
- Anthropic has not mounted comparable legislative campaigns around voter priorities like job displacement, data center impacts, or chatbot effects on children; Fernandez said the company is "eager to engage" on those issues but has produced no coordinated state-level pushes on them.
Why it matters: Anthropic's push for tougher state AI laws comes as critics from David Sacks to open-source researchers accuse it of regulatory capture. State legislators in Illinois, Massachusetts, California, and New York must now weigh whether Anthropic-backed bills genuinely address catastrophic risks or mainly raise compliance costs for competitors — right as the Trump administration forced the company to suspend its own most powerful models.
