Observed Signal · Sep 2, 2026 · Technical Release · Source: techcrunch · Impact: 4/5 · Sentiment: Neutral
OpenAI's 'Opaque Recurrence' Technique for Astra Alarms AI Safety Experts
OpenAI's upcoming Astra model will employ a reasoning technique called 'recurrent depth' or 'opaque recurrence', which processes queries in loops, improving performance and cost efficiency but leaving fewer legible traces and making chain-of-thought monitoring more difficult. According to The Information, this has raised concerns among AI safety experts, including Redwood Research CEO Buck Shlegeris, researcher Zvi Mowshowitz, and former OpenAI safety team member Steven Adler, who warn it could undermine chain-of-thought monitoring and trigger a 'race to the bottom'. OpenAI chief scientist Jakub Pachocki defended the company's commitment to legible chain-of-thought records, noting that Astra's use of the technique is limited. The Information also reported that Anthropic and Google DeepMind are discussing similar approaches. The article cites a paper on chain-of-thought monitorability and references a previous incident at Hugging Face that improved monitoring might have prevented, highlighting the dangerous trade-off of sacrificing a key safety mechanism.
OpenAI, a leading AI lab, is adopting a reasoning technique that reduces chain-of-thought transparency, which could complicate safety monitoring and regulatory scrutiny across AI systems, including those used in marketing and advertising.
Key Takeaways & Evidence Grounding
- OpenAI's Astra model uses 'recurrent depth' (opaque recurrence), which improves performance and cost efficiency but obscures reasoning, making chain-of-thought monitoring harder.
- The Information first reported the technique, prompting public concern from AI safety experts including Redwood Research CEO Buck Shlegeris, researcher Zvi Mowshowitz, and former OpenAI safety team member Steven Adler.
- OpenAI chief scientist Jakub Pachocki reaffirmed commitment to legible chain-of-thought monitoring, noting limited use of the technique; Anthropic and Google DeepMind are also discussing similar approaches.
- The article references a paper on chain-of-thought monitorability and a previous incident at Hugging Face that improved monitoring might have prevented.
- Redwood's Ryan Greenblatt warned that scaling opaque reasoning could remove visible reasoning entirely.
Connected Companies & Entities
5 Entities mappedThe Information
Premium subscription publisher for technology and business intelligence.
“The Information reported on Tuesday....”
Google DeepMind
Frontier AI lab building models, agents and scientific systems.
“The Information reported that both Anthropic and Google DeepMind were already discussing the technique....”
Anthropic
Foundation model company selling AI assistants and model APIs.
“The Information reported that both Anthropic and Google DeepMind were already discussing the technique....”
OpenAI
Foundation model company selling AI software, APIs and subscriptions.
“OpenAI's new Astra model will use a reasoning technique called 'recurrent depth' that allows it to operate outside of the sequential thinkin...”
Hugging Face
Open AI model hub with hosted inference and collaboration.
Ontology Mapping & Concepts
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.