Anthropic commits another $20 million to AI policy group

Anthropic said on July 21, 2026, that it is giving another $20 million to Public First Action, bringing its total contributions to the nonpartisan organization to $40 million. The company outlined the move in Anthropic News, saying the money is restricted to public education and policy work—not election activity involving federal, state, or local candidates.

Public First Action works with Republicans, Democrats, and Independents on AI safeguards. Anthropic says its goal is to build support for policies that preserve U.S. leadership while requiring stronger transparency and protections from developers of advanced models.

Why AI builders should care

The company’s argument is that capability growth is moving faster than governments’ ability to respond. Anthropic pointed to Claude Mythos Preview, which it says identified thousands of high-severity software vulnerabilities across every major operating system and browser. Rather than release the model broadly, Anthropic made it available to selected cyber defenders through Project Glasswing.

That example frames the policy questions facing AI tool builders: how should powerful systems be tested, who verifies safety claims, and what happens when a model creates unacceptable risks? Anthropic’s proposed framework calls for independent evaluations, public reporting of catastrophic-risk findings, enforceable safety practices, and the possibility of slowing or blocking dangerous deployments.

The company also backs tighter controls on advanced chips and semiconductor equipment, along with measures against illicit model access and distillation. For developers, those proposals could eventually shape evaluation requirements, security programs, deployment timelines, and access to frontier capabilities. Anthropic described the donation as one step toward making those debates more prominent, rather than a finished policy blueprint.

Source: Anthropic News