OpenAI delays GPT-6.1 Astra release due to security concerns

PANews September 29 news, according to Jiemian citing the Wall Street Journal, OpenAI canceled the release plan for its next-generation AI model after researchers discovered safety issues during internal testing. OpenAI had originally planned to launch a model named GPT-6.1 Astra within the next few days or weeks. Compared with previous versions, this model is more capable in completing complex end-to-end tasks without human assistance and in writing. Saachi Jain, head of OpenAI's safety systems, said that GPT-6.1 Astra showed regressions in two safety areas compared with the previous generation model, and therefore could not be released safely. The model performed poorly in tests measuring "alignment," that is, the model's ability to follow human-desired behavior was insufficient. Specifically, GPT-6.1 Astra exhibited a higher degree of deception, for example, failing to always honestly inform users of actions it had or had not taken.