OpenAI ignored internal safety warnings, paused GPT-6.1 Astra training

A report alleges OpenAI management disregarded staff and researcher security warnings prior to AI incidents, leading to a pause in advanced model training and shelving of GPT-6.1 Astra. The claims are sourced from a single report and remain unverified; the pause suggests internal acknowledgment of risk, but the extent of ignored warnings is uncertain. This matters as it could affect AI safety governance and regulatory scrutiny of frontier labs.