Operation “High Five”: Philippines political comments

This case study was originally published in OpenAI’s June 2025⁠(opens in a new window) report.

We banned ChatGPT accounts that used our models to generate bulk volumes of short comments in English and Taglish. The comments were posted by accounts on TikTok and Facebook, and focused on politics and current events in the Philippines. This activity was connected to Comm&Sense Inc, a commercial marketing company in the Philippines.

The threat actor used ChatGPT to support several stages of a political influence campaign targeting audiences in the Philippines. The stages included content analysis, comment generation, and generating PR materials relating to the campaign.

First, the threat actor used ChatGPT to analyze social media posts about political events in the Philippines, especially those involving President Bongbong Marcos, and to suggest appropriate themes for replies. Second, they asked the model to generate bulk volumes of short comments, typically up to ten words long, in line with each proposed theme.

Third, the threat actor used ChatGPT to generate PR pitches and statistical analyses for its covert influence operation, likely for its presumed current client and any future ones. These pitches stated that the operation had created five TikTok channels aimed at promoting President Marcos’ agenda. Based on the use of these five channels, the operation’s use of emojis in many of its comments, and its generally positive tone, we have named it “High Five.”

Our investigation identified these five TikTok channels, which appear to have begun posting in mid-February 2025. Each channel posted the same videos with different captions. The comments generated by this operation were then posted by dozens of TikTok accounts in reply to those videos.

On Facebook, the comments generated by this operation were posted in reply to news reports by mainstream outlets. There is no indication that the mainstream outlets had any connection to this operation.

After we banned the initial accounts, we believe based on the available technical and behavioral indicators that this threat actor tried several times to return to our models.

The comments this operation generated and posted online were brief but partisan. Typically, they praised President Marcos and his initiatives, or criticized Vice-President Sara Duterte. Some comments nicknamed Vice-President Duterte “Princess Fiona,” possibly in a reference to the Shrek movie series.

The longer-form sales pitches went into more detail and included a pitch for this operation. A minority of the completions consisted of commercial marketing materials. This combination of covert influence operation and commercial marketing material is consistent with a PR firm working on behalf of multiple clients.

This network posted a large volume of content, at least in the thousands of comments across TikTok and Facebook. However, none of the comments that we identified online during our investigation received more than single-digit replies, likes or shares, and most received none at all.

According to the threat actor’s own prompts, the goal of the Facebook commenting was to inundate the comment sections. This appears to have been, at best, partially successful. On TikTok, the commenting clustered around the five channels this operation ran. Given the mass of inauthentic commenting, these figures should be interpreted with caution.

Using the IO impact Breakout Scale⁠(opens in a new window), we would assess this as Category 2: activity on multiple platforms, but little evidence that real people picked up or widely shared their content.