Additional $20 Million Donation to Public First Action

Anthropic has made an additional $20 million donation to the nonpartisan AI policy organization Public First Action. Following its first donation in February 2026, the cumulative total support has grown to $40 million. These funds may only be used for public education and policy activities, and cannot be used to support election candidates.

Anthropic cited the rapid advancement of AI capabilities as the background for the additional donation. Earlier this year, Claude Mythos Preview discovered thousands of high-risk software vulnerabilities across all major operating systems and browsers, which Anthropic disclosed only to trusted cybersecurity defense experts through Project Glasswing.

As policy directions, Anthropic proposed ▲mandatory transparency requirements for AI companies ▲independent evaluation and civil penalties for risky models ▲authority to block deployment of models posing catastrophic harm risks, and has formalized these into the Anthropic Advanced AI Framework (AAIF). Anthropic also stated that it supports strengthened export controls on advanced chips and blocking illegal access to models in order to maintain AI leadership among democratic nations.