Operation “Sneer Review”: China-origin influence activity

OpenAI has pulled the plug on a cluster of accounts it says were using its models to mass-produce social media posts attacking a Taiwanese influencer, alongside a scattering of US-focused political content. The company tagged the operation with the codename Sneer Review, and pointed to China as the likely origin.

What's notable here isn't the scale, which by OpenAI's own account was modest, but the pattern. This is one of several influence campaigns the company has now disclosed where the actors weren't hacking anything or breaking new ground technically. They were just using ChatGPT the way a marketing team might, to churn out variations on a message faster than a human writer could, then spreading that content across platforms to make a fringe opinion look like a groundswell.

Taiwan has been a recurring target in these disclosures, and that's not surprising given the political stakes Beijing has in shaping opinion both on the island and among the diaspora audiences who follow influencers covering it. Attacking a single content creator might sound like small potatoes next to election interference or state propaganda, but reputational attacks on independent voices are exactly the kind of low-cost, deniable pressure campaign that's hard to trace back to a government and easy to repeat at scale.

OpenAI's response, as with prior takedowns, was to ban the accounts and publish what it found, rather than staying quiet. That's become the company's playbook for these disclosures: not dramatic, but it does mean the public gets a running tally of how state-linked actors are actually using generative AI in the wild, rather than just speculation about what they might do.