Anthropic AI model delayed over risks

Anthropic CEO Dario Amodei confirmed that the company is in direct talks with the US government to collaborate on mitigating AI-related national security risks.

Instead of a general release, the model will be accessible only to a select group of tech partners and government agencies through "Project Glasswing." This decision comes as a federal appeals court refused to block the Pentagon from labeling Anthropic a supply-chain risk.

The move highlights a growing trend of "radical transparency" regarding the potential dangers of advanced reasoning models in the tech industry.