Self-regulation as Hail Mary. OpenAI fires safety researchers. AI as the killer app for the US Department of Defense. Scientific misconduct questions, again. Indirect prompt injections.