AI Research3 min reading time
Anthropic wants tests, not bans, as OpenAI and Google back open weights
Covered by 3 sources
Read full postAnthropic CEO Dario Amodei opposes banning open-weight AI models and instead advocates for mandatory safety testing of sufficiently capable models before release. His proposal includes tighter export controls on advanced chips and cracking down on large-scale model distillation, aiming to balance innovation with safety. The industry lacks consensus on defining 'sufficiently capable,' which will determine which models require testing and regulation.


