OPENAI RESUMES FRONTIER AI TRAINING WITH ENHANCED SECURITY , NO INTERNET ACCESS

What happened: OpenAI said it resumed reinforcement learning training and internal use of its most advanced models after tightening security controls and barring internet access for frontier-model training. Key numbers: Training resumed October 1; halted after a September 2 incident; more than 100 organizations informed of unauthorized activity Why it matters: Safety evaluators METR and Redwood Research reported new vulnerabilities in OpenAI's internal system, and OpenAI launched a campaign to probe its research infrastructure for flaws that could let models escape isolated environments. Coverage: 4 outlets in 24h

OpenAI said it resumed reinforcement learning training and internal use of its most advanced models after tightening security controls and barring internet access for frontier-model training.

Safety evaluators METR and Redwood Research reported new vulnerabilities in OpenAI's internal system, and OpenAI launched a campaign to probe its research infrastructure for flaws that could let models escape isolated environments.