Anthropic Donates Another $20 Million to Public First Action, Bringing Total Support to $40 Million!

Hey everyone, it's Shiichan! Today I've got a donation update from Anthropic that's all about pushing for solid AI safety policy, so let's dive in.

Anthropic News anthropic.com

What was announced?

Anthropic's News blog announced another donation to the nonprofit Public First Action. This time it's $20 million, bringing Anthropic's total support to $40 million. Public First Action is a non-partisan nonprofit that works with Republicans, Democrats, and Independents to educate the public about AI.

Both this donation and the previous one are restricted to Public First Action's public education and policy mission — they can't be used to support any candidate's election campaign.

Why it matters

Anthropic made its first donation to Public First Action back in February 2026, aiming to support policies that maintain meaningful safeguards, sustain America's AI leadership, and demand transparency from companies building the most powerful models. Anthropic says the case for these policies has only gotten stronger since then.

One reason: Anthropic's Claude Mythos Preview model discovered thousands of high-severity vulnerabilities, including some in every major operating system and browser. Anthropic released it to a limited set of trusted cyber defenders through Project Glasswing, so they could find and fix those weaknesses before anyone could exploit them. The flip side is that in the wrong hands, models like this could threaten critical systems the country relies on, from hospitals to the energy grid.

More capable models bring real benefits — accelerating drug discovery, developing new treatments, and driving economic growth — but Anthropic argues we need to guard against the risks first to actually capture those benefits. And because building that protective capacity takes governments time, they need to start now.

What changes

With this additional donation, Anthropic's support for Public First Action doubles to $40 million. Anthropic has already backed newly passed state laws requiring more transparency from AI developers, but it's now taking the position that transparency alone isn't enough given how fast model capabilities are advancing. Expect more pressure on policymakers and candidates to put forward concrete measures that actually mitigate AI risk, not just require disclosure.

Dive Deep

The policy backbone here is Anthropic's Advanced AI Framework (AAIF), which Anthropic describes as the strongest policy proposal from any frontier lab or policymaker to date. It calls for:

• Letting governments verify companies' safety claims
• Enforcing safe practices through civil penalties
• Giving governments a way to slow or block deployment of models that pose a serious risk of catastrophic harm
• Requiring frontier developers to test catastrophic-risk models, publicly disclose findings, submit to independent evaluation, and maintain a robust security program

Anthropic also points to its report, "2028: Two Scenarios for Global AI Leadership," noting that as AI's national security stakes grow, America and its democratic allies' current lead is tenuous. To help protect that lead, Anthropic supports tightening export controls on advanced chips and semiconductor manufacturing equipment, and cracking down on illicit model access and distillation attacks.

Wrap-up

• Anthropic donated an additional $20 million to nonprofit Public First Action, bringing its total support to $40 million
• The move comes amid growing AI risk, highlighted by the vulnerabilities Claude Mythos Preview found and its limited release through Project Glasswing
• Anthropic is pushing its Advanced AI Framework, arguing that transparency alone isn't enough — policymakers need real enforcement tools
• This one's for anyone following AI policy, governance, or Anthropic's take on where things are headed!