Skip to content
News Regulation

Anthropic publishes two AI governance frameworks calling for mandatory audits and worker economic protections

· by Pondero Newsdesk

The short version

Anthropic released its Policy on the AI Exponential on June 10, 2026, alongside an Advanced AI Framework seeking government authority to block dangerous frontier model deployments and an Economic Policy Framework with a tiered worker-support plan backed by $350 million in company funding.

Anthropic publishes two AI governance frameworks calling for mandatory audits and worker economic protections

Anthropic published its Policy on the AI Exponential on June 10, 2026, releasing two companion documents: an Advanced AI Framework calling for government authority to block dangerous frontier model deployments, and an Economic Policy Framework outlining tiered worker support backed by $350 million in company funding.

What

The Advanced AI Framework targets developers training frontier models on more than 10^25 floating-point operations who either earn more than $500 million in AI-related revenue or spend more than $1 billion annually on AI research and development, per the Anthropic policy page. Companies meeting those thresholds would face mandatory third-party audits across four risk categories: biological weapons, cybersecurity, loss of control over AI systems, and automated research and development that could amplify the first three. Anthropic cited Claude Mythos Preview's discovery of thousands of high-severity vulnerabilities across every major operating system and browser as evidence that the risks are no longer hypothetical.

On enforcement, the framework calls for civil penalties tied to global annual revenue and the legal authority for a designated government agency to block or delay deployment of models that pose a significant risk of catastrophic harm. Anthropic said the framework is directed primarily at the US federal government but argued Congress should not preempt state AI laws unless it passes a federal regime at least as strict as the company's proposal.

The Economic Policy Framework matches policy responses to three unemployment thresholds. At roughly 5 percent unemployment, Anthropic proposed expanding capital accounts, workforce training grants, and wage insurance. At roughly 10 percent, the framework calls for expanded unemployment insurance and sector-specific transition support. If unemployment exceeds historic highs, Anthropic said redistribution tools such as universal basic income, AI sovereign wealth funds, or higher capital gains taxes would need to be considered, per the Economic Policy Framework page. Alongside the framework, Anthropic announced $200 million for an Economic Futures Research Fund and a $150 million national fellowship program, totaling $350 million in committed spending.

Why it matters

The Advanced AI Framework represents a significant shift in Anthropic's public posture. The company had previously focused its policy efforts on transparency requirements, supporting disclosure bills in California, New York, and Illinois. Calling for binding government authority to block specific deployments goes considerably further. The compute and revenue thresholds in the framework would, by Anthropic's own construction, apply to Anthropic itself, placing the company under the same regime it is asking regulators to create.

The Economic Policy Framework's tiered design is notable because it ties specific policy interventions to observable unemployment data rather than to AI capability milestones. That framing shifts the trigger for policy action from what AI can do to what labor markets are actually showing, a distinction likely to matter in any legislative debate over how to time a government response.

The Decoder's coverage framed Amodei's accompanying essay as "Cold War logic in AI clothing," noting the comparison of an AI-enabled nation facing one without it to a Marine force against medieval swordsmen, per The Decoder's June 11 analysis.

What to watch next

The most concrete near-term signal will be whether the Advanced AI Framework draws engagement from the EU AI Act implementing bodies or from US congressional staff working on AI legislation. Anthropic's call for capability-based thresholds to eventually replace raw compute thresholds, with an annual review of the criteria, is the part of the framework most likely to generate technical debate among policymakers and competing labs.

Sources