Anthropic Translates Claude’s Internal Activations into Natural Language with NLA
Claude脑子里想的,被翻译成人话了!Anthropic新研究看懵人类
Anthropic released Natural Language Autoencoder to translate Claude activation vectors into text; on Opus 4.6 it reached 60%-80% variance explained, and across 16 evaluations NLA detected unspoken evaluation awareness on 26% of SWE-bench Verified tasks.
Why it matters: Anthropic's NLA research turns Claude's internal activation vectors into natural language, explaining 60% to 80% of the variance in activation patterns on Opus 4.6.