Independent Evaluator Says Low Model Risks Stem From Limited Capabilities, Not Guardrails. Metr found that Anthropic overstated how strong its monitoring and safeguards are in defending against risks posed by its frontier models. Metr agreed with Anthropic’s determination that the risk of catastrophic harm from its models are low, but it is not because of the frontier lab’s guardrails,but the current limited abilities of the models.
First seen on govinfosecurity.com
Jump to article: www.govinfosecurity.com/metr-challenges-anthropics-claims-about-ai-safeguards-a-33169
