Anthropic is removing its model evaluation datasets from public internet access. This move aims to prevent external entities from using these datasets to train their own models, thereby protecting Anthropic's competitive edge. The company plans to make these evaluations available again in the future through controlled channels. AI
IMPACT Restricting access to evaluation data may slow down independent benchmarking and could impact the pace of AI development across the industry.
RANK_REASON Company is restricting access to its internal evaluation data, which is a product/tooling decision, not a core model release or research breakthrough.
Read on Mastodon — mastodon.social →
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →