Lesser Wrong Feed

 Takes One to Know One - Training a model to grade reward hacks causes it to reward hack less itself
 Tue, 06 Oct 2026 02:20:39 GMT
 You can graft SDF changes from base models onto post-trained models
 Tue, 06 Oct 2026 02:17:28 GMT
 Kurzgesagt's video on the Hugging Face incident: "AI Just Crossed the Terrifying Line - Now What?"
 Tue, 06 Oct 2026 01:29:30 GMT
 Three types of AI risk: why (some) leftists are against (some) AI regulations
 Tue, 06 Oct 2026 00:49:35 GMT
 A closer look at what ozone and nuclear arms control tell us about regulating AI
 Tue, 06 Oct 2026 00:46:53 GMT
 Humans Are Not Chinchillas: Revisiting “How Quick and Big Would A Software Intelligence Explosion Be?”
 Mon, 05 Oct 2026 23:41:10 GMT
 The Unacknowledgable Labor of Pastor's Wives
 Mon, 05 Oct 2026 23:29:41 GMT
 Dance, Dance
 Mon, 05 Oct 2026 23:05:44 GMT
 How Would We Know? Reflections on trying to make AI go well amid deep uncertainty
 Mon, 05 Oct 2026 19:58:12 GMT
 Self-Modeling Interventions Modulate Emergent Misalignment
 Mon, 05 Oct 2026 19:05:54 GMT