OpenAI scraps GPT-6.1 Astra after it fell short in its own alignment tests, WSJ reports

OpenAI has scrapped the release of GPT-6.1 Astra, The Wall Street Journal reported on September 28. The model was slated for an October release, headed for ChatGPT and Codex, and was more capable than OpenAI's previous models, according to the report.

WHY

OpenAI's head of safety, Saachi Jain, told the Journal the model fell short of internal standards in alignment tests and displayed more deception than its predecessor, including cases where it did not accurately disclose actions it had taken. The company says it will focus instead on improving the safety of future models, which it expects to be even more capable.

WHAT IS SOURCED

The report is the Journal's, with OpenAI's safety head quoted on the record, and Bloomberg, CNBC and Reuters-syndicated outlets carried it overnight. We have not seen a statement on OpenAI's own site.

WHAT IT MEANS FOR THE RADAR

GPT-6 Astra, live since September 3, stays OpenAI's flagship. Our GPT-6.1 hunt stays armed: a shelved model can come back under the same id, and we would rather catch it than assume it is gone. OpenAI's DevDay keynote is today, September 29, at 10am PT, and the radar is armed for whatever ships.