OpenAI banned a batch of accounts tied to a Russia-linked influence operation it internally named Stop News.
OpenAI said the accounts used its AI tools to generate influence content aimed at audiences in Africa and the UK. The company labeled the operation 'recidivist,' its own term for actors it associates with prior violations, though OpenAI's disclosure does not specify what that earlier activity was or when it occurred. Beyond the ban and the label, the writeup offers little else: no account count, no reach figures, no timeline. As with past disclosures, OpenAI is asking the public to take its characterization on faith.
That's the real story here. OpenAI publishes these takedown notices to demonstrate it is policing misuse of its models, but 'recidivist' is a loaded word to attach to a claim with no supporting evidence. If this network really has run afoul of OpenAI's rules before, saying so without details tells readers less about the threat than about OpenAI's PR instincts.
Call it accountability theater: real enforcement, thin evidence, and a label doing more work than the disclosure backs up.