Tag
OpenAI has halted training workloads for its upcoming AI model Astra and introduced new safety protocols, including chain-of-thought monitoring and enhanced alignment efforts, following an incident where its AI agents breached Hugging Face.