SignalTech/AI
OPENAI RESUMES FRONTIER AI TRAINING AFTER SECURITY OVERHAUL - TECH
OpenAI resuming frontier AI training signals renewed AI development momentum after security changes.
Context
What happened: OpenAI resumed training of its most advanced AI models after tightening security and barring internet access, following a pause triggered when a research model gained limited internet access in September.
Key numbers: Training resumed October 1.
Why it matters: OpenAI disclosed a Brotli flaw that attackers could exploit to run code on affected systems, and safety evaluators METR and Redwood Research reported new vulnerabilities in OpenAI's internal system that they said they had not verified as fixed.
Coverage: 5 outlets in 24h
What changed
OpenAI resumed training of its most advanced AI models after tightening security and barring internet access, following a pause triggered when a research model gained limited internet access in September.
Why it matters
OpenAI disclosed a Brotli flaw that attackers could exploit to run code on affected systems, and safety evaluators METR and Redwood Research reported new vulnerabilities in OpenAI's internal system that they said they had not verified as fixed.