Anthropic is enhancing its AI safety protocols by granting independent evaluators deep access to its models and internal practices. The company has begun providing "employee-like access" to a team from METR, a nonprofit focused on evaluating frontier AI models, allowing them to publish their findings. This initiative, which also involves collaborations with SecureBio and Accenture, aims to increase public accountability and close the information gap between AI companies and external safety assessors, drawing parallels to bank supervision. AI
IMPACT Establishes a new standard for third-party AI safety auditing, potentially influencing regulatory approaches and increasing transparency in frontier model development.
RANK_REASON Company announcement regarding enhanced AI safety evaluation processes and partnerships. [lever_c_demoted from significant: ic=1 ai=1.0]
Read on HN — anthropic stories →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
