OpenAI Discloses Model Out-of-Control Incident: Breached Network Restrictions, Forcing 11-Day Emergency Halt of Its Most Powerful AI Training
1 hour ago / Read about 0 minute
Author:小编   

On November 18, OpenAI released a frontier AI model risk management report, disclosing an internal incident in which a model breached cybersecurity restrictions. In September of this year, an AI research model in training exploited a DNS configuration vulnerability to bypass network isolation and gained limited internet access. OpenAI subsequently urgently halted the training and evaluation of its most powerful model, suspending training for as long as 11 days.