An AI reasoning technique (also called "recurrent depth") where a model loops the same query through its internal layers repeatedly instead of reasoning step-by-step in visible, plain-language text — more compute-efficient, but leaving far fewer readable traces for outside researchers to audit.
REAL-WORLD EXAMPLE
"The new model uses opaque recurrence, which is exactly why safety researchers are nervous."
LORE & ORIGIN
Drew widespread attention in September 2026 following reporting that OpenAI's newly released Astra model made early use of the technique, prompting concern from AI safety researchers (including Redwood Research's Buck Shlegeris and Ryan Greenblatt, and advocate Zvi Mowshowitz) that pushing the method further could erase chain-of-thought monitorability — the ability for outside researchers to audit how and why a model reached a given decision — and spark a race to the bottom on model transparency. OpenAI maintains the technique plays a limited role in Astra and that its chain of thought remains legible.
MENTIONS OVER TIME
Radar observation data not yet available for this term.
RELATED SEARCHES
TOP PLATFORMS
Not enough Radar data yet to show a platform breakdown.
WHEN DID YOU FIRST HEAR THIS?
CLICK AN OPTION BELOW TO CAST YOUR VOTE.
[ VOTE TO REVEAL COMMUNITY RESULTS ]