Lesser Wrong Feed

 Initial DIY Cleanroom Experimentation
 Tue, 22 Sep 2026 14:10:58 GMT
 Modern LLMs have tiny GPTs hidden inside them
 Tue, 22 Sep 2026 13:33:10 GMT
 Controllable-CoT leads to covert reasoning capabilities
 Tue, 22 Sep 2026 13:04:11 GMT
 Politics Gets Interested In Those Trying Not To Die
 Tue, 22 Sep 2026 13:00:53 GMT
 Autonomous Evidence Factories: Safe and Useful Recursive Self-Improvement
 Tue, 22 Sep 2026 12:28:32 GMT
 Projecting AI Automation at Anthropic
 Tue, 22 Sep 2026 11:29:25 GMT
 Do we agree what  "short timelines" means?
 Tue, 22 Sep 2026 10:16:27 GMT
 Total Safety Transparency?
 Tue, 22 Sep 2026 05:39:39 GMT
 Math contains a lot of stuff
 Tue, 22 Sep 2026 05:39:12 GMT
 When talked into harm, a model blames the answer, not itself (an interpretability study of guilt vs shame)
 Tue, 22 Sep 2026 04:43:09 GMT