The reliance on undisclosed sources in AI evaluations raises concerns about transparency and accountability in assessing potential risks.

The post Anthropic uses undisclosed source in its AI research capability evaluations appeared first on Crypto Briefing.