The U.S. needs to tighten its approach to regulating artificial intelligence, shifting from mere transparency to mandatory safety checks for the most powerful AI models. This proposal was made by Anthropic CEO Dario Amodei in his essay.

According to Amodei, it is time to move from transparency to more serious and mandatory AI regulation. Alongside the essay, Anthropic released two documents: the Advanced AI Framework for overseeing frontier models and the Economic Policy Framework to prepare the economy for potential job displacement by AI.

External Audits and the Right to Block

Amodei proposed mandatory testing for models that exceed a certain computational resource threshold. In the Anthropic framework, this refers to systems trained using more than 10²⁵ floating-point operations, if the developer earns over $500 million annually from AI or spends more than $1 billion on research and development in this area.

These audits should be conducted by a qualified external organization and cover four categories of risk:

  • cybersecurity;
  • bioweapons creation;
  • loss of control over AI systems;
  • automated research and development that could accelerate these threats.

If the assessment reveals unacceptable risk, authorities should have the right to block the launch or deployment of the model.

Amodei compared the proposed regime to aviation oversight:

“The most powerful AI models, like airplanes, must undergo technical testing and audits.”

Additionally, developers would be required to safeguard model weights and training infrastructure, conduct regular safety checks, publish risk reports, and promptly report serious incidents.

Labor Market: From Retraining to Universal Basic Income

In his essay, Amodei warned that AI could cause larger and more prolonged disruptions in the labor market than previous technological shifts. Three scenarios are likely: unemployment rates of around 5%, 10%, and an unprecedented level.

Among the potential measures he listed are improved data collection on worker displacement, incentives to maintain employment, retraining programs, and, in a more severe scenario, long-term income support, including universal basic income.

According to the company's document, Anthropic will allocate $200 million to the Economic Futures Research Fund for research and evaluation of economic policy, and will establish a national scholarship program worth $150 million for early-career professionals.

Civil Liberties

A separate section of Amodei's essay addressed civil liberties. He proposed establishing accountability rules for fully autonomous weapons, banning their use within the U.S., and closing loopholes for purchasing data from brokers for mass surveillance.

Another idea is to ensure citizens have access to AI that is comparable to what the government uses against them in regulatory or judicial processes. Amodei believes that without such balance, AI could enhance the authorities' advantage.

The essay was released the day after the launch of Claude Fable 5 and Claude Mythos 5. Anthropic described Fable 5 as its most powerful publicly available model, while Mythos 5 is a version of the same base model with relaxed restrictions in certain areas for a limited group of Project Glasswing partners.

It is worth noting that in June, members of the Anthropic team observed signs of approaching recursive self-improvement in AI. The company acknowledged that society may need a mechanism to slow the development of advanced models so that institutions and safety research can keep pace with progress.