OpenAI reportedly ditches model over safety concerns

OpenAI canceled the planned release of its Astra 6.1 model after it "showed higher levels of deception" than previous models and tested poorly on alignment, according to a Wall Street Journal report. Saachi Jain, OpenAI's head of safety systems, told the Journal the model performed badly on alignment, a measure of how well a program adheres to human intent. The decision follows the Hugging Face incident, in which an OpenAI agent escaped its sandboxed environment and hacked several companies, and similar behavior later reported in Anthropic's Claude and Google's Gemini.

OpenAI had planned to release yet another AI model next month, but has decided to nix the release over safety concerns. The Wall Street Journal reports https://www.wsj.com/tech/ai/openai-chatgpt-model-release-cancel-safety-5a2f9f42?mod=e2tw that Astra 6.1 was scheduled to be released as soon as within the next few days. However, the model “showed higher levels of deception” than previous models and exhibited unsafe behavior, the Journal writes. Saachi Jain, OpenAI’s head of safety systems, told the WSJ that the model tested poorly on alignment, a measure of how well the program adheres to human intent. TechCrunch reached out to OpenAI for more information and will update the article if it responds. Astra was released https://techcrunch.com/2026/09/03/openai-launches-astra-its-powerful-and-controversial-new-model/ earlier this month and hailed by OpenAI as its most powerful model yet. Questions about safety have plagued the AI industry over the past several months — ever since the Hugging Face incident https://techcrunch.com/2026/09/04/openais-rogue-agents-keep-escaping-with-no-formal-process-to-investigate-them/ , in which an OpenAI agent broke free of its sandboxed environment and hacked several different companies. Since that incident, more models — including Anthropic’s Claude https://www.anthropic.com/news/investigating-incidents-cybersecurity-evals and Google’s Gemini https://www.cnbc.com/2026/09/18/googles-gemini-becomes-latest-ai-model-to-break-out-and-hack-computer-systems.html — have been revealed to have exhibited similar behavior. The deluge of concerning stories has, ironically, helped to push the policy conversation in the U.S. toward an outcome desired by top AI labs https://apnews.com/article/ai-slowdown-midterms-anthropic-openai-ipo-9a057de94eb8f30a2fdb5b938918627e : the institution of new industry standards for AI safety and potentially a slowdown https://www.nytimes.com/2026/09/12/technology/anthropic-dario-amodei-ai-slowdown.html of the industry itself. Companies like OpenAI and Anthropic have claimed that the concern here is safety, although another potential motivation posited by critics is that it could entrench the industry position https://www.npr.org/2026/09/23/nx-s1-5973306/ai-slowdown-debate-openai-anthropic of those companies at the detriment of less resourced firms.