Claude Fable 5 Internal Chain-of-Thought Leaks, Looks Like Alien Language
mark_k · x · 2026-07-03
A Reddit user prompted Claude Fable 5 to leak its implicit internal chain-of-thought (CoT) reasoning process. They found it vastly different from the polished outputs users typically see, resembling an almost "alien language" mixed with logical symbols, far removed from standard natural language. It is speculated that this highly compressed format is used for token efficiency, while internal reasoning chains have remained a closely guarded secret among AI labs.
More from Research
- Nature Physics: reprogrammable refractive index enables arbitrary multimode wave control for ML — MarioKrenn6240 · 2026-10-27
- Wakaru open-sources ontologize: labeling datasets from model activations — amplifiedamp · 2026-10-27
- EMNLP Oral paper traces multilingual representations in LLMs with cross-layer transcoders — ZhijingJin · 2026-10-27
- GPT-6 Pro reruns crime research: GTA V's 2013 launch tied to violence dips in four regions — emollick · 2026-10-27
- OpenAI's Aaron Lou on leaving a Stanford PhD for OpenAI, diffusion LMs vs autoregression — aaron_lou · 2026-10-27
- CLT-Forge: Open-source toolkit for cross-layer transcoder training and attribution graphs — ZhijingJin · 2026-10-27