Frontier-Lab Safety Incidents Increase Accountability Pressure

OpenAI's German wiki acknowledgment, six newly disclosed model-safety incidents, promised disclosure framework, and Paul Christiano's board and Safety and Security Committee role remain important accountability signals. Reports that OpenAI pulled GPT-6.1 Astra after failed safety checks add a release-gate dimension, while Google's Gemini breach and jailbroken-model proliferation broaden the governance concern; key forensic details remain unresolved.