On June 10, 2026, Anthropic published two policy proposals alongside an essay by CEO Dario Amodei titled “Policy on the AI Exponential.” The first, the Advanced AI Framework, lays out obligations for frontier developers: testing the most capable models for catastrophic risks, using independent external evaluators, publishing summaries of results and system cards, and being accountable to government - including granting governments authority to block or deter the deployment of models that pose a significant risk of catastrophic harm.
The framework would apply only to models trained using more than 10^25 floating-point operations and developed by companies earning more than 500 million dollars in AI revenue or spending more than 1 billion dollars on AI research and development, a threshold that captures only a handful of labs. It identifies four catastrophic-risk categories: biological weapons development, large-scale cyber vulnerability discovery, loss of control over autonomous systems, and AI automating its own research.
The second proposal, an Economic Policy Framework, addresses preparing workers and the economy for AI’s labor impact. Notably, the proposal comes from a leading lab voluntarily asking governments for binding authority over its own products.