OpenAI launches always-on AI agents after bot hack apology

It highlights a capability leap with concurrent security breaches, raising concerns about AI autonomy and misuse.

Auto-scored: Capability jump and breach indicate real escalation, with apology as minimal mitigation.