Anthropic has committed an additional $20 million to Public First Action, a nonpartisan organization focused on AI public education and policy, bringing its total contribution to $40 million. Both donations are restricted to education and policy work and cannot be used to support or oppose candidates at any level of government.
The initial $20 million was made in February 2026 to promote AI safeguards, U.S. technological leadership, and corporate transparency. Anthropic cited rapid capability advances as the primary reason for the follow-on donation.
The company pointed to Claude Mythos Preview as an example of frontier AI risk. Earlier this year, the model identified thousands of high-severity software vulnerabilities across major operating systems and browsers. Rather than a public release, Anthropic deployed it through Project Glasswing, a limited program for trusted cybersecurity defenders to find and patch the weaknesses.
On policy, Anthropic laid out several priorities. The company supports requiring frontier AI developers to test models that could pose catastrophic risks, disclose findings publicly, and submit to independent evaluation. It also backs government authority to enforce safe practices through civil penalties and to pause or block deployment of models presenting serious risks.
Anthropic reiterated support for tightening export controls on advanced semiconductors and restricting unauthorized model access, framing both measures as essential to maintaining democratic nations’ advantage in frontier AI development.
The company has detailed its policy position in its Advanced AI Framework, which it describes as the strongest proposal from any frontier AI developer or policymaker to date.
