Research · Two Minute Papers ·
Claude AI Knows More Than It Tells You
Anthropic published a paper on natural language autoencoders and related interpretability work for Claude. The post links to the paper and a companion technical article, with no additional results described in the raw summary.