safetyopinionbearishEach new detail learned about the OpenAI incident makes the situation seem worse, suggesting serious incident response problemsZvi Mowshowitz27 Jul 2026https://thezvi.substack.com/p/more-on-an-internal-openai-model