While the US Congress continues to stall on federal AI regulation, Anthropic, the nearly trillion-dollar startup, is pushing states to adopt tougher laws. Cesar Fernandez, Anthropic's head of US state and local government relations, told WIRED that the transparency-focused safety bills passed in California and New York in 2025 are just a start. As AI capabilities advance rapidly, self-reporting alone is no longer sufficient for the most powerful systems.
The paradox of a pro-regulation giant
Anthropic's stance may seem odd for a company building a massive business on advanced AI. Yet, as previously reported, Anthropic's founders believe that to fulfill their mission of safely navigating the transition to transformative AI, they must build a large enterprise while simultaneously supporting rigorous regulation. The company has already endorsed measures in Illinois requiring third-party audits for AI labs, and more recently a Massachusetts bill that would empower the state attorney general to seek injunctive relief against non-compliant companies.
Sponsored Protocol
Accusations of regulatory capture and Anthropic's response
Not everyone applauds. David Sacks, former White House AI czar, has accused Anthropic of pursuing a regulatory capture strategy, pushing for laws so complex that smaller startups are trapped, thus securing Anthropic's dominant position. Fernandez flatly denies this, noting that the laws Anthropic supports apply only to large AI model developers with revenues exceeding $500 million. In practice, few startups meet that threshold. However, competitors like Safe Superintelligence, Thinking Machines Lab, and Mistral could soon qualify, raising legitimate concerns.
Fernandez insists the goal is to inspire a race to the top in developing safe AI systems, with transparency about risks to critical infrastructure as a top priority. This stance aligns with Anthropic's history: the company has long warned about catastrophic AI risks, and its GPT-Red automated red-teaming tool demonstrates its commitment to safety.
Sponsored Protocol
The limits of state intervention according to Anthropic
Despite pushing for stricter regulations, Anthropic believes the power to block unsafe model deployments should remain with the federal government, not states. This position is ironic given the recent Trump administration directive to suspend access to Anthropic's most powerful models for foreign nationals. The company criticized that move, calling for a fair and transparent evaluation process. At the federal level, Anthropic also accused Alibaba of a distillation attack, reigniting the open-source debate. Fernandez clarifies that at the state level, they have not targeted open-source models, but rather the capabilities of systems above certain thresholds.
Sponsored Protocol
While Anthropic focuses on catastrophic risks, many American voters are more concerned about job displacement, data center impacts, and chatbots' effects on children. The company promises to engage on these issues, but so far its legislative campaigns have neglected them. Whether Anthropic's influence will lead to a balanced approach or, as critics fear, entrench the power of a few giants remains to be seen. For further reading, see the analysis on Meta's manual review of teen chats.
For a comprehensive overview of AI regulation challenges, refer to the original WIRED article (source).
Source: https://www.wired.com/story/why-anthropic-is-pushing-states-to-regulate-ai-faster
