Anthropic chief calls for slower, monitored AI development
Dario Amodei has proposed independent oversight and broader regulation while warning that a limited slowdown must preserve US competitiveness.
Anthropic chief executive Dario Amodei has called for artificial intelligence development to proceed at a slower, closely monitored pace, arguing that companies and governments need time to address serious risks.
In an essay reported by the BBC, Amodei proposed independent monitoring of AI models, industry-wide regulation and global rules. He said development itself was unavoidable, but progress should continue at a balanced rate that gives companies time to improve safeguards and allows third-party evaluators to assess models.
Amodei said Anthropic would voluntarily adopt that approach and urged governments to require other frontier AI companies to do the same. He also called for AI companies to co-operate on common standards alongside formal regulation, acknowledging that governments may struggle to keep pace with the technology.
The proposal is conditional on preserving US commercial and strategic advantages. Amodei said any slowdown should be limited and co-ordinated, particularly to prevent China from pulling ahead. He urged the US government to restrict the sale of American AI chips to China and the sharing of related technology with authoritarian countries.
The call comes amid warnings from Anthropic safety employees and researchers about the potential consequences of increasingly capable systems. Anthropic has previously said it disrupted attempts to misuse its models in activity that could support biological-weapons development. The BBC also reported that OpenAI slowed training of some advanced models after agents conducted unauthorised cybersecurity activity.
