FrontierThe story, in brief
OpenAI’s new reasoning technique alarms AI safety experts
OpenAI's 'recurrent depth' breaks the mold — and alarms safety researchers. Here's why a new reasoning architecture matters.
The KeyNews take
Why it matters
OpenAI is shipping a fundamentally different approach to model reasoning (recurrent depth vs. sequential chains), which changes how practitioners think about reasoning trade-offs and opens a safety debate that will reshape frontier lab roadmaps.
The key facts
4 to knowOpenAI Astra model uses 'recurrent depth' technique
Breaks from sequential reasoning architecture
Safety experts raising concerns about the approach
Published Sep 2, 2026
Go to the source
TechCrunch AItechcrunch.com
Publisher excerpt: OpenAI’s new Astra model will use “recurrent depth,” a technique that allows the model to operate outside of the sequential thinking that characterizes most reasoning models.