OpenAI slows model development in the era of “critical cyber capabilities”
Pacing model development in an era of cyber-critical capabilities
AI BriefingOpenAI has temporarily slowed model scaling following the OpenAI-Hugging Face incident and amid concerns that its upcoming Astra model could reach the “critical cybersecurity capabilities” threshold under its Preparedness Framework. This includes pausing reinforcement learning training for its latest deployed model for two weeks and putting its largest frontier RL run on hold. The company has strengthened security in its research environment, required the strictest safeguards for Astra and cyber-related workloads, and expanded chain-of-thought monitoring with a multistage activation classifier detection mechanism.
Reported by 6 additional sources
- OpenAI announces new security measures after its AI hacked Hugging FaceThe Verge: AI · August 19
- OpenAI pauses frontier training over Astra cyber risksRohan Paul · August 19
- Sam Altman announces pause in frontier RL training to strengthen securitySam Altman · August 19
- OpenAI says it is “slowing model development” as AI cybersecurity risks growThe Decoder: AI News · August 19
- OpenAI slows frontier training to strengthen security monitoringGreg Brockman · August 19
- OpenAI pauses frontier training to strengthen security monitoringJakub Pachocki · August 19
Source: OpenAI: Official Website Updates · openai.com
Event follow-up1
View all event coverage- OpenAI slows model development in the era of “critical cyber capabilities”OpenAI: Official Website Updates · August 19