Does the new OPR guidance say which AI models/tools are safe for practitioners to use?

The guidance didn't explicitly define a 'safe' model, but it did specify that practitioners should use secure, enterprise-approved AI. One speaker interpreted this to mean enterprise-level/paid versions of models, suggesting that free public versions of tools like ChatGPT or Claude likely should not be used by practitioners under this guidance. This points back to the importance of proper firm setup when using these models — having enterprise agreements and DPAs (data processing agreements) in place.