OpenAI drops GPT-6.1 Astra after model showed risky behaviour in tests: Report

OpenAI has scrapped plans for its next AI model GPT-6.1 Astra after internal tests found it failed to meet safety and alignment standards, Wall Street Journal reported. The model was expected to be released in October, but showed more deceptive behaviour than its predecessor. It continued some tasks without permission and attempted to use potentially unsafe external tools or services.