Bottom Line: Anthropic calls for an aviation-like regulatory authority or commissioned private auditors to examine AI models for critical risks before their release.
Dario Amodei, CEO of Anthropic, demands state control mechanisms for powerful AI systems. Governments should be able to stop dangerous models, particularly when cybersecurity, bioweapons and loss of control are at stake.
Dario Amodei has outlined concrete demands for AI regulation in a blog post: powerful AI models must be reviewed for risks in advance before deployment. The focus would be on cybersecurity, bioweapons abuse, and potential loss of control over autonomous systems. Automated software advancement that could amplify these risks should also be subject to oversight.
Amodei cites an authority similar to state air traffic control as an organizational model. Alternatively, government-commissioned private auditors could also perform this task. With this, Anthropic positions itself as a proponent of stronger regulation—a stance that has met with some criticism within the AI industry and from the White House.
The background for this position is also Anthropic’s own research findings: the company’s Mithril AI model was able to uncover vulnerabilities in widely used software that had in some cases gone undetected for decades. This reinforces concerns that AI systems could be misused for large-scale cyberattacks. Anthropic, with its Claude language model, is among the direct competitors of OpenAI’s ChatGPT.
Source: www.it-daily.net · Published June 11, 2026
Lumi AI News — AI-assisted curation in accordance with Article 50 EU AI Act. Paraphrasing and classification by Lumi News Pipeline v1.6.5.