Close Menu
Tech Savvyed
  • Home
  • News
  • Artificial Intelligence
  • Gadgets
  • Apps
  • Mobile
  • Gaming
  • Accessories
  • More
    • Web Stories
    • Spotlight
    • Press Release

Subscribe to Updates

Get the latest tech news and updates directly to your inbox.

What's On
What happens when AI detectors fail? Researchers say we must be trained to spot fake AI faces

What happens when AI detectors fail? Researchers say we must be trained to spot fake AI faces

12 July 2026
Your next Spotify song could soon carry an AI warning label, and the music industry is all for it

Your next Spotify song could soon carry an AI warning label, and the music industry is all for it

12 July 2026
Volkswagen’s ID. Unyx 09 doesn’t look like any VW I’ve seen, and I want it in the US

Volkswagen’s ID. Unyx 09 doesn’t look like any VW I’ve seen, and I want it in the US

11 July 2026
Facebook X (Twitter) Instagram
Facebook X (Twitter) Instagram
Tech Savvyed
SUBSCRIBE
  • Home
  • News
  • Artificial Intelligence
  • Gadgets
  • Apps
  • Mobile
  • Gaming
  • Accessories
  • More
    • Web Stories
    • Spotlight
    • Press Release
Tech Savvyed
Home»News»What happens when AI detectors fail? Researchers say we must be trained to spot fake AI faces
News

What happens when AI detectors fail? Researchers say we must be trained to spot fake AI faces

News RoomBy News Room12 July 20263 Mins Read
What happens when AI detectors fail? Researchers say we must be trained to spot fake AI faces
Share
Facebook Twitter Reddit Telegram Pinterest Email

Artificial intelligence has become remarkably good at creating fake human faces. So good, in fact, that the old tricks people relied on – counting fingers, spotting warped earrings, or looking for distorted backgrounds – are quickly becoming obsolete. According to a new study highlighted by the BBC, the next line of defence may not be a better AI detector at all. It might simply be a better-trained human.

Researchers from the University of Aberdeen, working alongside Australia’s National University, found that people can dramatically improve their ability to distinguish AI-generated faces from real ones after a relatively short period of structured training. Instead of hunting for obvious visual glitches, participants were taught to recognise subtle patterns that modern image generators still struggle to replicate consistently.

The AI race is forcing humans to evolve too

For years, identifying AI-generated images felt almost trivial. Early models often produced six fingers, mismatched earrings or impossible shadows. But today’s generators, powered by systems such as StyleGAN3 and newer diffusion models, have largely moved beyond those tell-tale mistakes. As a result, researchers argue that relying on visual defects is no longer an effective strategy.

Instead, participants were trained to judge six perceptual qualities that AI faces often share. These include unusually perfect facial symmetry, highly proportional features, above-average attractiveness, generic-looking facial structures, limited emotional expression, and faces that are surprisingly difficult to remember after you’ve looked away.

The results were striking. Before training, participants correctly identified AI-generated faces only around 40 percent of the time. After roughly an hour of guided learning and repeated exposure to both real and synthetic faces, accuracy climbed to nearly 80 percent. A handful of participants even approached perfect detection scores. More importantly, their confidence became better aligned with their actual performance, something earlier research suggested was often missing.

Why spotting AI faces matters more than ever

This isn’t simply an academic exercise anymore. Deepfake technology is already being used in financial fraud, political influence campaigns and online identity scams. The BBC points to Deloitte estimates suggesting losses from AI-enabled deepfake fraud in the United States could rise to £40 billion next year, up sharply from around £12 billion in 2023. It also references a widely reported Hong Kong case in which scammers allegedly used a deepfake video call to convince an employee to transfer £25 million. Meanwhile, an earlier Associated Press investigation uncovered an AI-generated LinkedIn profile that successfully infiltrated US policy circles.

The study also highlights another important issue: AI systems remain less reliable at generating older faces, younger faces and people from underrepresented ethnic groups because of biases in their training data. Those imperfections may still provide useful clues for human observers.

Artificial Intelligence

Perhaps the most interesting takeaway is that the human brain appears to learn much like AI itself. By repeatedly seeing examples of real and fake faces, people gradually develop an intuitive sense of authenticity rather than relying on a single giveaway. Researchers believe that instinct may become one of our strongest tools as generative AI continues to improve.

The irony is difficult to ignore. As artificial intelligence becomes better at pretending to be human, humans may have to start training themselves the way machines do – through data, repetition, and pattern recognition. AI detectors may keep improving, but the research suggests they shouldn’t be the only defence. Human judgement still has a role to play; it just needs an upgrade.

Share. Facebook Twitter Pinterest LinkedIn Telegram Reddit Email
Previous ArticleYour next Spotify song could soon carry an AI warning label, and the music industry is all for it

Related Articles

Your next Spotify song could soon carry an AI warning label, and the music industry is all for it

Your next Spotify song could soon carry an AI warning label, and the music industry is all for it

12 July 2026
Volkswagen’s ID. Unyx 09 doesn’t look like any VW I’ve seen, and I want it in the US

Volkswagen’s ID. Unyx 09 doesn’t look like any VW I’ve seen, and I want it in the US

11 July 2026
China’s answer to SpaceX’s reusable rockets literally catches boosters in a net

China’s answer to SpaceX’s reusable rockets literally catches boosters in a net

11 July 2026
Galaxy Watch 9 and Ultra 2 may arrive on July 22, and this new leak leaves little to the imagination

Galaxy Watch 9 and Ultra 2 may arrive on July 22, and this new leak leaves little to the imagination

11 July 2026
Instagram is letting people generate AI images of you. Here’s how to stop it.

Instagram is letting people generate AI images of you. Here’s how to stop it.

11 July 2026
Disney+ is exploring a free tier to fight back against YouTube’s growing TV dominance

Disney+ is exploring a free tier to fight back against YouTube’s growing TV dominance

11 July 2026
Demo
Stay In Touch
  • Facebook
  • Twitter
  • Pinterest
  • Instagram
  • YouTube
  • Vimeo
Don't Miss
Your next Spotify song could soon carry an AI warning label, and the music industry is all for it

Your next Spotify song could soon carry an AI warning label, and the music industry is all for it

By News Room12 July 2026

The music industry’s battle with artificial intelligence is entering a new phase. After spending the…

Volkswagen’s ID. Unyx 09 doesn’t look like any VW I’ve seen, and I want it in the US

Volkswagen’s ID. Unyx 09 doesn’t look like any VW I’ve seen, and I want it in the US

11 July 2026
China’s answer to SpaceX’s reusable rockets literally catches boosters in a net

China’s answer to SpaceX’s reusable rockets literally catches boosters in a net

11 July 2026
Galaxy Watch 9 and Ultra 2 may arrive on July 22, and this new leak leaves little to the imagination

Galaxy Watch 9 and Ultra 2 may arrive on July 22, and this new leak leaves little to the imagination

11 July 2026
Tech Savvyed
Facebook X (Twitter) Instagram Pinterest
  • Privacy Policy
  • Terms of use
  • Advertise
  • Contact
© 2026 Tech Savvyed. All Rights Reserved.

Type above and press Enter to search. Press Esc to cancel.