Anthropic Calls for Mandatory Safety Tests on Powerful Open-Weight AI Models

Pressing for strict oversight, Anthropic demands mandatory safety testing for powerful open-weight AI models, but what hidden risks are driving this urgent push?