Especially because the data that gets fed into training is first anonymized, so they’d need to look for navier stokes related stuff in the anonymized training s

Especially because the data that gets fed into training is first anonymized, so they’d need to look for navier stokes related stuff in the anonymized training set and then get make some sort of ad hoc process (with Tristan’s permission and sign off from legal) to compare the training data against his chats / Codex sessions to check if anything matches up. And that assumes his chats / sessions are still there, and not deleted to compare against.