Anthropic is contributing an additional $20 million to Public First Action, bringing its total support for the nonpartisan organization to $40 million. The funding is designated for public education and policy work related to artificial intelligence.
Anthropic said the money cannot be used to support or oppose candidates for federal, state or local office. The company previously donated $20 million to Public First Action in February 2026.
Public First Action works with Republicans, Democrats and independents to promote public understanding of AI and advocate for safeguards governing advanced systems. Anthropic said the organization’s activities align with its support for transparency, safety requirements and continued U.S. leadership in AI.
The company argued that the increasing capabilities of frontier AI models create potential economic and scientific benefits alongside cybersecurity and national security risks. Anthropic believes governments should begin developing the expertise and regulatory capacity needed to address those risks before more powerful systems are deployed broadly.
Anthropic cited the cybersecurity capabilities of Claude Mythos Preview, which the company said identified thousands of high-severity software vulnerabilities. Access to the system was limited to selected cyber defenders through Project Glasswing, enabling trusted organizations to identify and remediate weaknesses.
The company said advanced AI could accelerate scientific research, drug discovery and economic growth. However, it also warned that capable models could threaten critical infrastructure when used without appropriate safeguards.
Anthropic supports requirements for frontier AI developers to test models that could create catastrophic risks. Its policy position also calls for public disclosure of relevant findings, independent evaluations and robust security programs.
The company believes governments should have the authority to verify safety claims, impose civil penalties for failures and potentially slow or block the deployment of models presenting a serious risk of catastrophic harm.
Anthropic also supports tighter export controls covering advanced semiconductors and chipmaking equipment. It has advocated measures intended to prevent unauthorized access to powerful models and techniques that copy or extract their capabilities.
The company described its additional contribution to Public First Action as an effort to increase public and policymaker attention to these issues. Anthropic noted that its proposed policy framework represents its own position and remains part of a broader continuing debate over AI governance.
