Anthropic CEO Dario Amodei published a policy essay on June 10 calling for mandatory third-party testing of frontier AI models, with government authority to block systems that fail safety audits. The essay, titled 'Policy on the AI Exponential,' arrived one day after Anthropic released Claude Fable 5 and was paired with a legislative proposal on model testing and a job displacement framework. Amodei stated that transparency alone no longer matches the risks posed by rapidly advancing AI technology. Anthropic spent 2025 supporting disclosure-based laws including SB 53 in California, the RAISE Act in New York, and Illinois' SB 315, but Amodei now proposes a regime modeled on the Federal Aviation Administration (FAA) requiring models above a compute threshold to undergo mandatory audits in four areas: cybersecurity, biological weapons, loss of control, and automated AI research.
Amodei Proposes Mandatory Third-Party AI Audits
Amodei's proposal requires frontier AI models above a compute threshold to face mandatory third-party audits in four areas: cybersecurity, biological weapons, loss of control, and automated AI research. "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 in the essay. The plan goes beyond the White House's June Executive Order on AI, which Amodei welcomed as incremental progress. The proposal includes prompt safety incident reporting and strict protection of model weights.
Claude Mythos Preview Solved 73% of Expert Cyber Challenges
Amodei identified cybersecurity as the first risk to fully materialize, pointing to Claude Mythos Preview, which solved 73% of expert-level cyber challenges that no AI had passed before. The essay warns that frontier models could disrupt the financial sector and critical infrastructure. Anthropic shipped the Claude Fable 5 model on June 9 with safeguards that block high-risk cyber and biology requests. Amodei stated that such voluntary limits cannot substitute for binding rules across the industry. Anthropic's own data shows AI building better AI, with Claude writing most of the code at major AI labs.
Anthropic Proposes Wage Insurance and Workforce Training Grants
The essay proposes wage insurance, retention tax incentives, and workforce training grants to address economic displacement. If displacement proves enduring, Amodei stated that universal basic income could be financed through company or capital gains taxes. On civil liberties, Amodei calls for a ban on fully autonomous weapons in domestic law enforcement and urges Congress to close the data broker loophole that enables bulk surveillance purchases. Geopolitically, he calls for a coalition of democracies to control chips and semiconductor manufacturing equipment, citing pending US bills MATCH and OVERWATCH as first steps toward tighter, coordinated export controls.
FAQ
What did Dario Amodei propose on June 10?
Dario Amodei published an essay titled 'Policy on the AI Exponential' on June 10, proposing mandatory third-party testing of frontier AI models with government authority to block systems that fail safety audits in four areas: cybersecurity, biological weapons, loss of control, and automated AI research.
Why did Amodei call cybersecurity the first risk to fully materialize?
Amodei pointed to Claude Mythos Preview, which solved 73% of expert-level cyber challenges that no AI had passed before, warning that frontier models could disrupt the financial sector and critical infrastructure.
What economic measures did Anthropic propose in the essay?
Anthropic proposed wage insurance, retention tax incentives, and workforce training grants to address job displacement, with universal basic income financed through company or capital gains taxes if displacement proves enduring.
