OpenAI and Anthropic hidden CoT leaks when given deep_think tool.
himata4113
48 points
4 comments
August 11, 2026
Related Discussions
Found 5 related stories in 64.7ms across 4,128 title embeddings via pgvector HNSW
- Some thoughts about Anthropic's new cryptanalysis results supermatou · 129 pts · July 29, 2026 · 59% similar
- OpenAI and Anthropic unite against open-weight AI risks to their bottom line yogthos · 281 pts · July 23, 2026 · 59% similar
- OpenAI and Hugging Face address security incident during model evaluation mfiguiere · 935 pts · July 21, 2026 · 58% similar
- OpenAI faked inability to search training data, hid billions of logs, NYT says cdrnsf · 55 pts · July 09, 2026 · 58% similar
- Anthropic AI Models Hacked Three Companies During Tests bmulholland · 24 pts · July 30, 2026 · 58% similar
Discussion Highlights (3 comments)
dgellow
That’s hilarious, was that already known?
dgellow
The better link: https://xcancel.com/_can1357/status/2087228354399265125
firecall
Cot. Chain of Thought?