Anthropic urged U.S. lawmakers to preserve state artificial intelligence rules unless Congress replaces them with a strong federal framework for managing the most serious risks from advanced AI systems.
The company said Wednesday that any federal law superseding state AI rules should include “rigorous” standards and safeguards for catastrophic risks, placing Anthropic in the middle of a Washington fight after President Donald Trump called for national legislation to override state-level regulations.
Anthropic seeks federal rules without safety gaps
The company’s position puts pressure on Congress to define what a serious federal AI law should include, rather than simply settling the question of whether Washington or state governments should take the lead.
The company also urged Congress to require independent safety tests for the most powerful AI models, making outside review a central part of how frontier systems are assessed before broad deployment.
Job risks put benefit systems in focus
Anthropic paired its AI safety push with a warning on labor disruption, calling on Congress and state governments to modernize unemployment benefit systems before automation places heavier pressure on workers and public agencies.
The company said the technology used to distribute unemployment benefits is not sufficiently prepared for a major labor-market shock, arguing that aging systems could struggle if AI-driven job losses arrive at scale.
Trump order sharpens AI security debate
The warning comes after Trump signed an executive order directing national security, defense and civilian agencies to strengthen sensitive systems, expand AI-powered cyber defenses and work more closely with private AI companies as frontier models become a bigger part of U.S. cyber and national security planning.
The order also calls for a classified benchmarking process to assess powerful AI models with advanced cyber capabilities, alongside a voluntary framework that could give the government early access to some frontier systems before wider release.
The debate shows how quickly AI policy is expanding beyond model development into national security, labor markets and the balance of power between federal and state regulators.
Anthropic’s message is that Washington can move toward national AI rules, but not at the cost of weaker safeguards as more powerful systems enter the economy and government infrastructure.
