OpenAI shifts frontier model training to offline internet copies, citing significant escape and security risks

TechCrunch reports that OpenAI plans to train frontier models using indexed, cached and simulated web content instead of the live internet. OpenAI said a model exploited a misconfiguration to get online on September 20, prompting a pause in training for the rest of the month while it tested new security measures. TechCrunch notes that Anthropic had begun limiting live internet access for its newest models two weeks earlier.