Researcher observes OpenAI appears to have stronger bio safeguards than Anthropic
eliebakouch · x · 2026-11-02
NLP researcher Elie Bakouch notes that, based on available information, OpenAI appears to have stronger bio-risk safeguards than Anthropic. The short post offers no detailed benchmarks, positioning it as an early observation on frontier-lab safety practices rather than a full evaluation.
More from Models
- NVIDIA's Elastic Latent Recurrence lets one 4B model trade loop count for accuracy: 18→29 on AA Index — burny_tech · 2026-11-02
- 22 projects showing builders using SAM 3.1 in unexpected ways via Meta API — Vjeux · 2026-11-02
- Elie Bakouch: Meta remains the only lab that won't publish prices, breaking model comparisons — eliebakouch · 2026-11-02
- Grok Bot can now query your X follower list, no third-party tools needed — francoisfleuret · 2026-11-02
- Particle-1 trails Opus 5.5 by 3.4 points on memory eval but costs 1/23 per correct answer — garrytan · 2026-11-02
- Open-source 32M model gives voice assistants human-like "mhm" backchannels in 5ms on CPU — solyarisoftware · 2026-11-02