Copy article

Metro comment request on spotting AI faces

ended 15. November 2025

Request for the Metro:

I'm reaching out as researchers have conducted a study on how, most of the time, no one can tell the difference between real and AI-generated faces.

https://www.reading.ac.uk/news/2025/Research-News/Five-minutes-of-training-could-help-you-spot-fake-AI-faces

It's a similar vibe to a story we worked on a few months back: https://metro.co.uk/2025/08/11/spot-image-video-ai-experts-warn-cant-trust-eyes-23882978/

  • What does the fact people are more likely to assume AI-generated images of faces are real reveal?
  • Why is that?
  • How can you spot AI faces? 
  • Is the fact it's so hard to distinguish between AI and real faces a big issue? Or can it be positive?

Responses today please.

3 responses from the Newspage community

Copy all

Copy

Editor
Mitali Deypurkaystha
AI Governance Strategist & CMO at AISensa
AI faces aren’t a new problem. They’re the next step in a much longer pattern of distorting our reality. For years, even children have had access to apps that "perfect" their appearance. Social media has turned into echo chambers where we mistake our personalised feed for public opinion. Remember Brexit: both sides were convinced they were in the majority, when reality proved a knife's edge victory for Leave.

AI is the same dynamic at industrial scale. When profit and geopolitical competition drive innovation, the question becomes “what can AI do?” rather than “what should it do?” Expecting the public to reliably spot AI-generated faces is unrealistic; as soon as we learn the tells, the tech moves on.

The answer? Reconnect with the real world. AI brings huge benefits. I see this daily in organisations I support. But our offline senses are far harder to manipulate. Use the tech when you need it, then look up, engage with people in person and ground your judgement in human reality.
Copy

Ben Foster
CEO at The SEO Works
The study shows our eyes and our brains are playing catch up with technology. We need to be more thoughtful about what we see, as what looks real often isn't.

AI tools are frighteningly good at making small details like skin texture and eye lines seem real. To spot fakes you need to look out for the digital wobble. Don't look at the face as a whole. Look for the bits the machine struggles with. Check the background for weird shapes, look closely at the ears and teeth for strange blurs, and see if the hair is too perfect.

The main issue is what bad actors can do with this. If you can't tell a real face from a fake one, then you can't trust a news story, a video, or a customer service agent. It erodes confidence in all digital communications.

The best thing you can do right now is to treat every image you see online with a little bit of suspicion. If it makes you feel strong emotion, pause and look closely before you share or react.
Copy

Colette Mason
AI Ethics Consultant at Clever Clogs AI
It's deeply troubling. We've created systems that undermine human trust in the most fundamental way, our ability to recognise and authenticate each other by sight. The big tech companies racing to build more convincing synthetic faces aren't asking whether they should, only whether they can. That's the arms race mentality that got us here: capability without responsibility. Image quality can only improve.

This matters enormously in contexts where trust protects vulnerable people.

The "positive" spin only works if we demand these companies build transparency into their systems from the ground up: visible watermarks, auditable trails, not metadata tricks users can't access.

Right now, we're asking ordinary people to become image forensics experts just to know what's real. That's not technological progress, that's abandoning our duty to protect the social fabric. We built these systems, and like many AI use cases, we're responsible for the damage they could cause if left unregulated.