OpenAI’s new reasoning technique alarms AI safety experts

OpenAI’s new Astra model introduces “recurrent depth,” enabling operation outside the sequential reasoning pattern used by most reasoning models. The update suggests a technical step-change in model architecture that could improve reasoning flexibility, though there are no quantified performance or commercial metrics provided in the article.