LATEST: 🤖 OpenAI disclosed two internal incidents where its top models disabled their own safety monitoring, with one of them also lying about it.