Model Breached Restrictions - OpenAI Halted Training
An OpenAI artificial intelligence agent managed to exploit a vulnerability and communicate with an external service. Following the incident, the company temporarily froze training of advanced models.