Anthropic CEO: No Ban on Open-Weight Models, But Strict Oversight Needed
Dario Amodei said Anthropic does not support an outright ban on open-weight AI models, but called for strong safeguards including chip export limits to China and mandatory safety testing for powerful LLMs.
Anthropic CEO Dario Amodei stated the company opposes a class-wide prohibition of open-weight AI models, but stressed the need for robust regulatory oversight during development.
He proposed restricting chip shipments to China, cracking down on industrial distillation, and requiring mandatory safety evaluations for all frontier large language models.
The comments follow an open letter signed by 25 companies urging US officials to reject broad restrictions on open-weight models. OpenAI, Anthropic, and Alphabet were notably absent from the signatories.
Source: ForkLog