Inside the TikTok network boosting ‘confessions’ of flood control corruption

Upgrade to High-Speed Internet for only ₱1499/month!

Enjoy up to 100 Mbps fiber broadband, perfect for browsing, streaming, and gaming.

Visit Suniway.ph to learn

Until its removal in November, hundreds of TikTok accounts had spent months magnifying the visibility of AI-manipulated videos of lawmakers appearing to admit to corruption in flood control projects. 

MANILA, Philippines — Hundreds of largely throwaway TikTok accounts have been amplifying a recently created channel that posts AI-manipulated videos of lawmakers appearing to admit to corruption in flood control projects, suggesting a coordinated inauthentic campaign that used spam comments to game TikTok’s algorithm and push fabricated political content to millions of users. 

The TikTok account, @bistado.files, known as “Gising Pampanga,” was no longer accessible as of November 5, shortly after Philstar.com reached out to TikTok for comment.

Between January and September 2025, Gising Pampanga uploaded 23 videos targeting former Pampanga third district representative Aurelio “Dong” Gonzales Jr., former House Speaker Martin Romualdez and their allies in the House. Nine used AI-generated deepfakes making them seemingly confess to pilfering flood control funds. One video reached 1.7 million views. 

Each upload triggered waves of low-effort comments from accounts — similarly worded insults and emoji chains — creating the illusion of organic, spontaneous outrage and pushing the videos onto TikTok’s “For You” page.

Before the account was deleted, the videos circulated freely on TikTok despite violating the platform's August 2025 policy banning unlabeled AI-generated content and deepfakes that depict public figures "linked to criminal behavior" or making statements on issues "they haven't actually addressed." 

The platform further warns that accounts engaging in spam or coordinated inauthentic behavior may be banned, restricted from appearing in search results or the For You feed, or have "fake likes, followers, or other inflated signals" removed. TikTok states it “strictly [prohibits] automation tools, scripts, or other tricks designed to bypass” its systems. 

TikTok's response. A representative of TikTok told Philstar.com in an email that the account has been permanently banned for violating the platform’s policies on AI-generated content. Philstar.com was told that TikTok requires creators to disclose realistic AI-generated or significantly edited media to prevent the spread of misleading information, and that the company enforces its community guidelines regardless of whether or not the content is labeled as AI-generated.

The representative added that the platform uses a combination of automated detection systems, human moderation, and user reports to identify and remove harmful synthetic media. TikTok employs moderators across languages and regions “to ensure appropriate enforcement in diverse local environments,” they said.

From April to June 2025, TikTok said 97% of AI-generated content removed for violations was taken down before users reported it. Still, the company acknowledged that detecting synthetic media at scale remains “an industry-wide challenge,” adding that it continues to update detection models and partner with experts to improve its accuracy.

Coordinated behavior

The coordinated activity around Gising Pampanga offers a snapshot of how TikTok’s engagement-driven system can be exploited to push politically motivated disinformation in local contexts, and how gaps in the platform’s moderation, particularly in non-English languages, allow such operations to thrive.

A Philstar.com analysis of over 11,000 comments posted across the account’s 23 videos revealed that more than half (57.6%) were duplicates or near-duplicates, indicative of scripted or prompted coordination. Aside from identical comments, the accounts used variations on core themes, sharing keywords and messaging while avoiding exact repetition to evade copy-pasta detection. 

Many comments either repeated short insults or offered low-effort agreements. Common keywords included “buwaya” (crocodile, slang for corrupt politician), “crocs,” “gising” (wake up), “tama” (correct), “ayos” (okay), “galing” (great), and “legit.” “Zero vote” and “never again” appeared frequently in videos posted ahead of the May midterm polls. 

Other comments hurled accusations like “ahas” (snake), “linta” (leech), “hudas” (traitor), “balimbing” (turncoat) and “epal” (attention-seeker). 

Nearly 43% of the comments were emoji-only, often three crocodiles or three laughing faces.   

At least 875 handles commented across two or more videos, forming a core cohort. One account appeared in 13 videos, a pattern that indicates coordinated multiposting. 

The dataset also included at least 370 disposable accounts, identifiable by usernames that follow a uniform “user[8-9 digit number]” pattern. 

Comment timelines show deliberate synchronization for the most viral videos. In four of the 23 videos, hundreds of comments appeared in tightly packed bursts or within the same minute. 

The account’s most-viewed video — a fabricated clip of former Gonzales calling himself the “flood control king of Pampanga” and Romualdez saying in Filipino “they don’t know we’re dividing up the funds among us” — logged more than 600 minute-level comment bursts.

No AI labels

TikTok requires creators to label AI-generated or significantly edited content that shows "realistic-looking scenes or people." Its August 2025 Community Guidelines prohibit "AIGC (AI-generated content) or significantly edited content that misleads about a matter of public importance," particularly content showing "a public figure being degraded, harassed, or linked to criminal behavior."

None of the nine AI-manipulated videos posted by Gising Pampanga carried the required labels. 

Interestingly, comment threads on these deepfaked videos also contained what appeared to be preemptive defensive messaging, lines such as "akala ko AI pero totoo pala" (I thought it was AI but it's real), "AI pero totoong nangyayari" (AI but it's really happening), and "hindi yan AI" (that's not AI). 

Cross-platform reach

The AI-manipulated videos were also cross-posted to other platforms and gained traction there. On Facebook, the page “Kasama sa pagbabago 2nd district of Pangasinan” reposted the video of the fake conversation between Gonzales and Romualdez about flood control funds. This drew over 300 reactions and over 260 shares as of posting. 

On X, account “@esoJngaM” reposted the same TikTok video, generating 290 likes and 96 reposts.