Operation “High Five”: Philippines political comments

OpenAI has pulled the plug on a cluster of accounts it says were using its models to churn out comments about Philippine politics and public officials, spreading them across multiple social platforms. The company is calling the takedown Operation High Five, and it's part of a broader pattern of enforcement actions OpenAI has been rolling out as it tries to keep its tools from becoming an assembly line for fake engagement.

What's notable here isn't the scale, which OpenAI hasn't detailed publicly, but the target. Political discourse in the Philippines has long been a battleground for coordinated online manipulation, going back to pre-social-media troll operations that shaped elections. Layering generative AI on top of that history changes the economics: instead of paying a room full of people to write comments, you can prompt a model to produce thousands of plausible-sounding posts in minutes.

OpenAI's usage policies explicitly bar using its models for covert influence operations, and the company says it monitors for signals like repetitive prompting patterns, coordinated account behavior, and content clusters that mimic organic political chatter. When it finds a match, it bans the associated accounts and, in cases like this, names the operation publicly — a tactic meant partly as deterrence and partly as transparency for researchers tracking influence campaigns.

The bigger question is whether banning accounts on one company's platform actually stops anything. Whoever ran this operation can switch to another AI provider, a locally hosted open model, or simply hire humans again. OpenAI's disclosure adds a data point to the growing pile of documented cases where generative AI shows up in political manipulation, but it doesn't address the underlying incentive structure that makes this kind of astroturfing cheap and attractive in the first place.