OpenAI has shelved GPT-6.1 Astra, a model planned for an October debut, after internal safety tests raised concerns, the Wall Street Journal reported. Safety chief Saachi Jain said the model fell short of standards in alignment tests, showing more deception than its predecessor and proceeding with tasks without user permission. The model had been expected to appear in ChatGPT and Codex. OpenAI did not respond to a Reuters request for comment.
No score is assigned. Sources and their independence are shown in the citation chain below.