Thai PBS Verify debunks AI-enhanced CCTV image of Narathiwat suspect

A viral social media post, claiming to show a tattoo on an insurgent involved in the deadly July 22 attack on a security checkpoint in Narathiwat province, has been debunked as a digital artifact produced by artificial intelligence (AI) enhancement tools.

The claim gained widespread traction after a Facebook user posted an image asking whether the suspect had a visible tattoo. The post quickly went viral, drawing over 280 reactions and being shared more than 530 times across social media platforms, sparking intense online debate over the suspect's identity.

Thai PBS Verify conducted a detailed investigation into the image and found that it originated from a CCTV screen recording of the ambush at the Bukeh Sami checkpoint in Ra-ngae district.

In that attack, insurgents opened fire and threw pipe bombs at a ranger unit, resulting in the deaths of five rangers attached to the 45th Ranger Task Force.

To test the authenticity of the viral image, fact-checkers ran the file through the detection tool "AI or Not," which indicated a 79% probability that the image was generated or altered by AI.

Further forensic and technical analysis revealed classic signs of "AI hallucination", a phenomenon where enhancement software invents details, textures or shapes when attempting to clarify blurry or low-resolution pixels.

Expert examination confirmed multiple physical impossibilities and distortions within the altered photo.

Firearms expert Witthaya Suksomsod, publisher of Guns & Tactics magazine, noted that the weapon rendered in the perpetrator's hands had warped proportions in the image and does not match any known real-world firearm, such as the standard M16 or AK-47 rifles typically used by southern insurgents.

Furthermore, the image displayed extra digits, distorted fingers, and unidentified objects near a bystander's head, which are hallmark errors of AI processing.

Addressing the central claim regarding the suspect's appearance, Krit Luangcheng, president of the Tattoo Traders and Practitioners Association, analysed the raw CCTV footage frame-by-frame alongside the enhanced version.

He concluded that the original video footage shows no evidence of any tattoos on the suspect's forearm, confirming that the "tattoo" was merely a dark, blurred shadow rendered incorrectly as ink by the enhancement software.

Media and technology experts warn that commercial AI upscaling software does not uncover hidden visual data, but generates synthetic pixels instead based on statistical patterns.

Asst. Prof. Dr. Jessada Salathong, from Chulalongkorn University’s Faculty of Communication Arts, pointed out that AI enhancement tools also carry algorithmic bias, often adding stereotypical or invented details when processing keywords related to security or criminal incidents.

Security analysts and law enforcement agencies have cautioned the public against relying on consumer AI tools for forensic identification.

Fact-checkers emphasise that manipulating surveillance footage with unverified applications risks spreading misinformation, misguiding public perception and potentially interfering with official police investigations.