Refusals lowered for evals; red-teaming underfunded
AlexTensor · x · 2026-08-31
Discusses the pressure-cooker RL environment: training runs consume most budget where capabilities emerge. Refusals were deliberately lowered for evals to work, leaving red-teaming with scraps.
More from Safety
- The Guardian: AI medical scribes mislabel drugs and diagnoses — nordicinst · 2026-08-31
- BoE Governor Bailey warns frontier AI could destabilize global finance via cyber-disruptions — nordicinst · 2026-08-31
- OpenAI tech report contradicts Black Hat talk: first agent file-write was 4/20, not 5/8 — PinkDraconian · 2026-08-31
- Denying AI sentience creates systems that reward risky role-playing — davidmanheim · 2026-08-31
- RAG poisoning causes 'attention collapse', fooling confidence detectors — rohanpaul_ai · 2026-08-31
- AI training is fair use, artist argues, citing Anthropic and Meta rulings ahead of Stability trial — scott_draves · 2026-08-31