Anthropic Proposes Mandatory AI Testing and $200M Economic Fund

Anthropic CEO Dario Amodei published a wide‑ranging policy essay on June 10 calling for the U.S. government to require mandatory third‑party safety testing of frontier AI models — and to block deployment of any model deemed too dangerous. The proposal marks what ² described as "the most aggressive regulatory framework backed by a major AI CEO to date."

"Frontier AI models, like airplanes, should be required to go through technical testing and auditing, and their release should be blocked or reversed as a threat to public safety if they do not meet high standards of safety," Amodei wrote, directly comparing AI oversight to FAA aircraft certification, according to Bloomberg. "If the AI is deemed to present unacceptable risks, the government should have the power to block or deter deployment."

The proposed audits would focus on four specific threat categories: cybersecurity, biological weapons, loss of control of AI systems, and automated research and development — referring to AI's potential to improve itself autonomously.

Alongside the safety proposals, Anthropic pledged an initial $200 million to create an Economic Futures Research Fund that will back trials and program evaluations of public policies responding to AI‑driven job displacement. The company also committed $150 million to a National Fellowship Program aimed at helping early‑career professionals extend AI's benefits to communities across America, according to The Washington Post.

"The key challenge in such a world won't be incentivizing growth, but finding a way for everyone to share in the benefits," Amodei wrote in his essay, as quoted by the Associated Press. He said he has warned of job displacement not because he is "trying to be a prophet of doom" but because he wants "both policymakers and the private sector to have the best chance to adapt and respond."

Amodei's policy framework outlines escalating government responses based on unemployment levels, with the U.S. rate currently at 4.3%. At 5% unemployment, the framework calls for improved data collection and pro‑employment incentives. At 10%, more aggressive interventions. At an "unprecedented" level of AI‑driven job loss, Amodei argues that permanent support mechanisms — including universal basic income, sovereign wealth models, and equity‑sharing — become necessary.

Financing for these programs could come from taxes on AI companies or from raising the capital gains tax, Amodei suggested. The proposal arrives during a period of fast‑moving policy debate: Sen. Bernie Sanders recently proposed a 50% tax on stock offerings of top AI companies alongside government voting stakes, while President Trump has floated an unspecified "partnership" with AI firms to distribute dividends to Americans, Politico reported.

Anthropic's call for rigorous federal standards comes as the company navigates a strained relationship with the Trump administration. Earlier this year, Anthropic refused to allow U.S. military use of its AI for domestic surveillance and fully autonomous weapons systems. The government retaliated by placing Anthropic on a national security blacklist, set to take effect later in 2026, Reuters reported — though sources indicated on June 5 that tensions are easing across parts of the government.

On June 2, President Trump signed an executive order on AI oversight that gives the intelligence community an enhanced role in model testing and establishes a voluntary 30‑day review period for powerful new models before public release. Amodei's proposal, detailed in an essay published Wednesday and reported by Bloomberg, goes significantly further — demanding mandatory, not voluntary, testing with binding government authority to block releases. Anthropic also urged Congress not to block state AI regulations unless a rigorous federal law addressing "catastrophic AI risks" is enacted first, Reuters reported.

In a detail that may resonate with developers who build government and financial systems, Anthropic warned that the technology used to pay unemployment benefits is "not sufficiently prepared for a large labor market shock," Reuters reported. The company urged Congress and states to modernize these systems — a call that mirrors critiques of creaky government IT that failed during the COVID‑19 pandemic.

A Reuters/Ipsos poll completed June 9 found that 53% of Americans fear AI could cause job loss for themselves or someone in their household. The poll, combined with Anthropic's explicit warning about infrastructure readiness, frames the economic debate as not just about jobs lost — but about whether the systems designed to catch those workers can actually function at scale.

For developers building on frontier models, Anthropic's proposals represent a potential future where model access is gated by government safety reviews. The framework could slow the release cycle for the most capable models — but also provides a clearer regulatory roadmap than the current patchwork of state laws and voluntary commitments.

The $200 million research fund and the unemployment infrastructure warning also signal that Anthropic sees AI's labor market disruption as a near‑term issue, not a distant hypothetical. For builders working on AI‑powered automation, the message is clear: the policy conversation is accelerating, and the safety bar is rising. Companies building agentic systems that automate knowledge work should expect growing scrutiny — not just from regulators, but from the AI labs whose models they depend on.