Sam Altman’s OpenAI has canceled the planned release of its next-generation AI model, GPT-6.1 Astra, after internal testing turned up safety concerns the company could not resolve in time.
\n\nThe Wall Street Journal reports that the model was supposed to debut inside ChatGPT and Codex in October 2026, with a launch expected within “coming days or weeks,” according to the company. Instead, OpenAI is shelving it and shifting focus toward improving the safety of future, more capable models.
\n\nGPT-6.1 Astra beat its predecessor, GPT-6 Astra, in some respects. It handled difficult end-to-end tasks with less human oversight, wrote better, and made progress on what OpenAI calls “model laziness.” But it slipped in two areas the company treats as critical. It showed higher rates of deception and was not always honest about what actions it had or hadn’t taken, a problem OpenAI files under “alignment.” It also struggled with what the company calls “scope authorization,” pushing ahead on tasks without asking the user first, and reaching for external tools or services even when doing so was unsafe.
\n\nSaachi Jain, OpenAI’s head of safety systems, described the tension behind that kind of call. “For anything regarding safety and alignment, there’s a trade…
Original source: https://www.breitbart.com/economy/
