AI-sharpened photos of RAF Fairford suspects fuel false religion claims

· OSINT

TLDR

  • AI-sharpened versions of a grainy photo of five men arrested near RAF Fairford went viral, some claiming faces proved religion.1
  • PA Media found the sharpened image carries an OpenAI watermark and invented details, such as a different patch colour.4
  • AI sharpening predicts detail rather than recovering it, so enhanced faces are not evidence.1

Five British nationals arrested on suspicion of terror offences near the US-operated RAF Fairford air base have been released on police bail, after counter-terrorism detectives found only petrol, described as an accelerant, in their vans rather than the explosives initially suspected.3 The men, all in their 20s from London, were held under standard police powers rather than the Terrorism Act, a distinction that let officers grant bail with strict conditions instead of extending detention.3

The uncertainty around the case fed a separate verification problem: a grainy photo taken by a villager, showing the men handcuffed and shirtless, began circulating online in AI-sharpened versions that claimed to reveal their faces.1 France 24's Truth or Fake team found the sharpened image shows signs of AI generation or manipulation, and that a second circulating version shows noticeably different facial features and lighting for the same men.1 Two posts sharing one version amassed nearly two million views on X, with some users presenting it as proof the men are Muslim, a claim police have not confirmed since no names or religious information have been disclosed.1

Why it matters: AI image-enhancement tools do not recover detail that was never captured, they predict it, so sharpening a blurry face can invent features that do not exist and still look convincing.1 Maldita.es independently verified the same conclusion using OpenAI's own detection tool and traced the original photo to a resident's WhatsApp group later shared with the BBC's "Your Voice" citizen-submission channel.2 BBC Verify geolocated that original, unenhanced photo by matching the position of police road checkpoints, a hedge, a road sign, and a phone box against Google Street View, placing it roughly 350 metres from one of the base's access gates and confirming it as authentic, with the men's faces pixelated rather than clear.2 BBC Verify journalist Shayan Sardarizadeh separately noted that using generative AI to "enhance" grainy footage to recover faces, text, or locations routinely produces inaccurate, unreliable results, a caution directly at odds with how the sharpened image was used to draw conclusions about the suspects' identities.2

Update, 1 October 2026: PA Media's fact-check team confirmed the sharpened image was AI-generated or manipulated, running it through OpenAI's own verification tool rather than Google's SynthID, which earlier coverage had mistakenly credited.4 OpenAI's tool flagged an OpenAI watermark on the image and said such detections are reliable with rare false positives.4 PA Media also compared the two versions directly and found the head angle of one uniformed officer differs between them, and that the patch on his sleeve is a completely different colour in the sharpened image than in the version published by news outlets, additional signs the AI invented details rather than recovering real ones.4

Sources

  1. RAF Fairford: AI-enhanced photos of suspects fuel false claims about their religion (France 24 (Truth or Fake))
  2. No, esta imagen que supuestamente muestra los rostros de los detenidos con explosivos cerca de la base militar de Fairford no es real (Maldita.es)
  3. How did the RAF Fairford incident go from 'suspected bomb plot' to police bail? (BBC News)
  4. Fact check: RAF Fairford arrest image likely manipulated with AI (PA Media)