OpenAI’s new reasoning technique alarms AI safety experts
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.
OpenAI’s Astra model introduces a technique called recurrent depth, and the reaction from safety researchers is immediate and uneasy. The method allows the model to loop through its own computations internally, rather than committing to a single forward pass of reasoning. In plain terms, the model can think in circles before it speaks, and that changes the calculus of oversight.
Read full analysis →