News

OpenAI’s Astra Reasoning Technique Raises AI Safety Concerns

Astra’s recurrent-depth reasoning is raising AI safety concerns as experts question whether increasingly hidden model reasoning could weaken monitoring and oversight.