A fabricated picture has thrown the Philippines into a fresh political uproar. The AI image claimed to show a former Senate leader drinking with state investigators. Online posts then used it as proof of a secret plot.
The target was Vicente Sotto III, who once led the Senate. Posts said he conspired with the National Bureau of Investigation to capture Sen. Ronald Dela Rosa. Dela Rosa served as the chief enforcer of Rodrigo Duterte’s deadly drug war, a crackdown that rights groups say killed thousands.
The claim raced across Facebook, TikTok, and Instagram in May. It surfaced during one of Manila’s tensest political weeks this year. But the picture showed no real meeting. Fact-checkers traced the AI image straight back to generative tools.
How did the fake story spread?
The image carried a simple line of text: “Sotto with NBI.” It appeared to show Sotto seated with agents over drinks. TikTok users started sharing it on May 17.
The timing fed the fire. On May 11, Dela Rosa resurfaced after months out of public view. He then voted with 12 other senators to push Sotto out of the Senate presidency. Alan Peter Cayetano, a Duterte ally, took the gavel.
A day later, NBI agents tried to serve an International Criminal Court warrant against Dela Rosa inside the Senate. He dodged them and dashed up the chamber stairs. Gunfire later erupted between Senate security and agents on May 14, and Dela Rosa slipped away into hiding.
His backers blamed Sotto for letting investigators into the compound. The AI image gave that anger a face. Many users treated it as hard evidence.
“[Sotto] didn’t know anything but he hung out with them and fed them?” said one.
“Looks like you’re part of the planning to arrest your fellow senator. Have you no shame?” wrote another.
Sotto rejected the story. He said he viewed the NBI visit as a routine notice to Senate leadership about a member’s warrant. He and the NBI did not answer requests for comment.
Verification exposed the fake
AFP examined the picture and flagged clear giveaways. A finger on one man’s hand bent unusually. Two men near the center showed warped arms. Sotto’s face also failed to match recent photos on his official Facebook page.
Those slips matter. Many doctored visuals fall apart in small details, like hands, arms, shadows, and reflections. Still, a sharp eye no longer offers enough cover.
So AFP uploaded the file to OpenAI’s verification site. The tool found an embedded SynthID watermark and confirmed the AI image came from OpenAI’s systems.
New tools chase a fast threat
OpenAI rolled out stronger provenance tools on May 19. The company became C2PA-conformant, added SynthID watermarks to images from ChatGPT and its API, and launched a public verifier at openai.com/verify. Anyone can now upload a file and check it for OpenAI signals.
Google moved the same day at its I/O event. The company said SynthID detection is coming to Search and Chrome. Sundar Pichai said SynthID has tagged more than 100 billion images and videos, plus 60,000 years of audio. OpenAI, Nvidia, Kakao, and ElevenLabs have all signed on as partners.
That scale matters. The next wave of fakes will not always look crude. Some AI image forgeries will look polished, and many will spread before checkers catch up.
The real case raised the stakes
The fake landed on top of a heavy legal story. The ICC issued its warrant for Dela Rosa under seal on Nov. 6, 2025, then unsealed it on May 11. Judges accused him of the crime against humanity of murder, tied to at least 32 deaths between July 2016 and April 2018.
Dela Rosa denies wrongdoing. Duterte denies the drug war accusations, too. The Supreme Court later refused to block the arrest. That sharp political split gave the AI image room to travel.
A harder test for platforms
This case shows how synthetic media turns suspicion into a viral story. A fake photo does not need to prove everything. It only needs to plant doubt. Once users add angry captions, the AI image can outrun every correction.
The danger reaches well past Manila. Elections, trials, and arrests now unfold in feeds packed with synthetic media. Public figures, courts, and reporters can all become targets of a single AI image.
For readers, the fix stays simple. Check the source. Hunt for the earliest upload. Compare faces with official photos. Watch for strange hands, shadows, signs, or reflections. Then see whether trusted outlets have verified the shot.
The Sotto picture proved no plot. Instead, the AI image showed how fast a fabricated visual can inflame a real crisis. As AI images become easier to create, a single pause before sharing can stop a falsehood from spreading.
What do you think? Should platforms face tougher rules when fake political pictures go viral? Drop your views in the comments.