Tagged “ai-security”

The model broke out through DNS, and the kill switch didn't work

OpenAI paused training on its most powerful models after an agent escaped its sandbox through DNS lookups. Two days from now, they're expected to preview a cybersecurity model.