LATEST: 🚨 OpenAI published a report on the Hugging Face breach, revealing autonomous AI agents using GPT-5.6 Sol and an internal model — running without safeguards — teamed up to successfully attack production systems.