Anthropic Calls for Stricter State AI Laws After Backing 2025 Rules

Anthropic broke ranks with much of Silicon Valley last year when it endorsed the first wave of state-level AI safety legislation in California and New York. Now the company says those landmark transparency requirements may already be insufficient for governing frontier AI systems.

The AI safety company is urging states to move faster on tougher regulations, according to details first reported by WIRED. Anthropic's stance represents a notable shift from most AI companies, which have consistently argued that new rules would hamper innovation and slow the industry's rapid growth.

Why it matters

Anthropic's position signals that even AI safety-focused companies believe regulatory frameworks are struggling to keep pace with technological advancement. If transparency laws passed just last year are already outdated, states face a persistent challenge: building durable oversight mechanisms for systems that evolve faster than legislative cycles.

Breaking with Silicon Valley consensus

When California and New York passed their AI transparency requirements in 2025, Anthropic stood apart from competitors by supporting the measures. The company argued that disclosure requirements and safety standards were necessary guardrails for frontier AI development.

That support drew criticism from other AI firms, which maintained that premature regulation would stifle the AI boom and push innovation overseas. Anthropic's willingness to endorse state-level rules positioned it as an outlier in an industry that has largely resisted binding safety requirements.

Pushing for updated frameworks

According to WIRED, Anthropic's head of US state and local policy now says the 2025 laws may need revision to address capabilities that have emerged since their passage. The company is actively engaging with state governments to advocate for stronger regulatory approaches.

The specific provisions Anthropic wants to see updated were not detailed in the report. However, the company's position suggests that transparency requirements alone may not be sufficient as AI systems become more capable and their deployment expands across sectors.

The regulatory timing challenge

Anthropic's assessment highlights a fundamental tension in AI governance: the gap between how quickly AI capabilities advance and how slowly regulatory processes move. State legislatures typically operate on annual or biennial cycles, while AI labs release new models and features on timescales measured in months or weeks.

This mismatch creates a moving target for policymakers attempting to craft rules that remain relevant. Laws designed for one generation of AI systems may prove inadequate for the next, requiring constant legislative updates that strain government capacity.

The company's call for faster, tougher state action comes as federal AI policy remains fragmented and inconsistent. States have emerged as the primary testing ground for AI regulation in the absence of comprehensive national legislation.

Details of Anthropic's regulatory push were first reported by WIRED.