OpenAI Flags New Model 'Astra' as Requiring Extra Safety Measures After Internal Testing
OpenAI announced on Sept 1 that an internally tested model named Astra exhibits capabilities well beyond those of the most advanced publicly available OpenAI model, GPT-5.6 Sol. The company said Astra's elevated performance means it will be developed and released with additional safety layers. The disclosure follows a separate incident in which Ope…