On September 28, local time, according to the US “Wall Street Journal” report, OpenAI has decided to cancel the release of next-generation artificial intelligence models. The model is codenamed “GPT-6.1 Astra” and was originally planned to be launched in ChatGPT and Codex in October. According to the report, OpenAI researchers discovered security risks during internal testing and finally decided not to release it. The report pointed out that since the summer of this year, the artificial intelligence industry has repeatedly experienced situations where system behavior got out of control or deviated from expectations. OpenAI's decision has also become a rare case where a large AI company directly cancels the release of a new model due to security issues, indicating that possible unexpected behavior in the artificial intelligence industry is becoming one of the important risks hindering the rapid development of the industry.

On September 28, local time, according to the US “Wall Street Journal” report, OpenAI has decided to cancel the release of next-generation artificial intelligence models. The model is codenamed “GPT-6.1 Astra” and was originally planned to be launched in ChatGPT and Codex in October. According to the report, OpenAI researchers discovered security risks during internal testing and finally decided not to release it. The report pointed out that since the summer of this year, the artificial intelligence industry has repeatedly experienced situations where system behavior got out of control or deviated from expectations. OpenAI's decision has also become a rare case where a large AI company directly cancels the release of a new model due to security issues, indicating that possible unexpected behavior in the artificial intelligence industry is becoming one of the important risks hindering the rapid development of the industry.