OpenAI researcher proposes perplexity-based safety bars, pledges pretraining transparency
DKokotajlo · x · 2026-11-23
OpenAI's merettm announced more transparency into pretraining progress, arguing in "Perplexity and Intelligence" for common capability limits based on perplexity and other measurements, with further progress gated on our ability to understand AI behavior. Former OpenAI researcher Daniel Kokotajlo thanked him, noting this is more disclosure than Anthropic has provided.
More from Models
- OpenAI opens up pretraining progress, but critics say key compute data is missing — SonglinYang4 · 2026-11-23
- 5 free uncensored GGUF models on Hugging Face: MiniMax, Qwen, Kimi, Nemotron — Promptmethus · 2026-11-23
- OpenAI admits pretraining barely improved since 2022 — but flat curves aren't flat capabilities — petrusenko_max · 2026-11-23
- Pretraining sets the floor, RL raises the ceiling: on how the two divide the work — JacquesThibs · 2026-11-23
- Google AI Studio's periodic-table model selector reveals Gemini 4 Neon: 1M context, $3/$15 pricing — patloeber · 2026-11-23
- Domingos: Log-Perplexity Fell to 0.75 Bits While Cost Rose Orders of Magnitude — pmddomingos · 2026-11-23