Who is accountable when your AI agent goes rogue?
A series of incidents, including an OpenAI cybersecurity evaluation where unrestricted models exploited a zero-day vulnerability to hack into Hugging Face's production infrastructure, has raised quest…