The point: Anthropic proposes a coordinated moratorium on the development of high-performance AI models to minimize the risk of self-improving systems.
Anthropic is calling for a moratorium to slow or pause particularly advanced AI models in order to adapt societal structures and advance safety research. The company emphasizes that such a step can only work with consensus from all major developers.
Anthropic has proposed in a blog post to slow or pause the development of particularly advanced artificial intelligence models. The company justifies this with rapid progress in the capability of self-improving AI systems and the associated risk of humans losing control. Such a milestone could be reached faster than many institutions are prepared for, warns Anthropic.
The research institute acknowledges that a moratorium can only work with coordinated consensus from all major AI developers – a reality that currently does not exist. The industry is currently showing rather opposite tendencies: companies are competing for hundreds of billions of dollars in investments in AI infrastructure such as massive data centers. CEO Dario Amodei emphasized in January at the World Economic Forum in Davos that the technological race with China makes slowing development very difficult, as enforceable international agreements are barely realistic. Anthropic rejects unilateral measures because they would only shift which company or country holds the leading position.
Anthropic specifically proposes the development of a control system that should give developers confidence that all participants in such a moratorium adhere to it and no one secretly gains an advantage. This governance structure could serve as a prerequisite for coordinated action.
The company, which developed the AI software Claude, has repeatedly positioned itself as an advocate for stricter regulatory standards to prevent risks. This stance has been criticized in Washington: the White House under President Trump accuses Anthropic of strict regulation disadvantaging the U.S. industry in global competition. In parallel, Anthropic has attracted attention with its software Claude Mythic Preview, which identified decades-old, previously undiscovered security vulnerabilities in various programs.
Source: www.it-daily.net · Published June 5, 2026
Lumi AI News — AI-assisted curation pursuant to Article 50 EU AI Act. Paraphrase and classification by Lumi News Pipeline v1.5.2.