For most of photography’s history, a picture was proof. If you saw it, it happened. That assumption is gone, and it isn’t coming back. AI image generators have gotten good enough that a fake photo and a real one can look identical, even to a trained eye. The tools that make them are cheap, fast and improving every few months. What used to take a Photoshop expert an afternoon now takes anyone with a phone about 10 seconds.

In July 2026, a fake photo showing Senator Mitch McConnell hooked up to hospital tubes spread across Reddit and X, timed perfectly to weeks of real silence from his office about an actual, unexplained absence. Fact checkers only caught it because Google’s SynthID watermark was still embedded in the file. A few months earlier, a fabricated photo appearing to show Bangladesh’s women’s cricket team wearing burqas after a World Cup loss spread across South Asia, until a wire service reporter confirmed no player had worn one and viewers spotted a stray Gemini AI logo in the corner. Neither fake was made by a spy agency or a criminal syndicate. Both worked because they landed at the exact moment people were primed to believe them.

That’s part of what makes this hard. Fake images come from teenagers experimenting, from scammers running romance or investment cons, from political operatives trying to sway an election and from people who just want to see what happens when a joke goes viral. The motives vary. The tools are the same, and they keep getting better.

For years, people were told to look for tells: hands with too many fingers, garbled text in the background, eyes that didn’t quite match, lighting that fell in two different directions. Those tips were useful advice for a while, and now they’re mostly out of date. The newer generators have fixed most of those errors. Looking for glitches isn’t a reliable strategy anymore because the software companies making these tools often compete with one another for the most sophisticated image generator.

Now Apple has entered the fray with a new feature on its iPhone 18 Pro and Pro Max called Reference Image, part of a system to spot fakes. Apple isn’t the first company to attempt this. A standard called C2PA, backed by Adobe, Microsoft and most major camera makers, has been around since 2022 and is meant to do something similar by tracking an image’s history from capture to publication. But C2PA has had security problems, including a flaw serious enough that Nikon revoked all of its C2PA certificates. 

With Apple Reference Image, when you shoot in this mode, the camera’s sensor assigns the raw image data with a secure cryptographic key, essentially sealing a fingerprint onto the photo at the moment of capture. Apple’s servers can later confirm whether an image matches its original sensor data, which means someone viewing the photo can check whether it was altered after the fact.

The feature works only on the two new phone models, and the verified images can only be viewed inside Apple’s own apps. A fake photo taken on any other device, or a real photo taken on an older iPhone, gets no such protection. 

It’s a step in the right direction, but reliably identifying doctored photos is still in the future. In the meantime, never take a photo at face value, especially the ones that show up in your social media feed. If a photo looks too dramatic or too perfectly timed to confirm what you already believed, that’s a good reason to not share it.

While major news outlets are held to editorial standards and are generally a safer bet than a random post, they aren’t immune either. News organizations have mistakenly run old photos, mislabeled images and gotten fooled by manipulated content before. 

If you’re in doubt about an image, do a reverse image search by uploading a screenshot into Google. Check whether more than one independent source is reporting the same image or event. And as verification tools like C2PA and Apple Reference Image become more common, look for those credentials on major platforms the way you’d check for a padlock icon on a website.

Leslie Meredith has been writing about technology for more than a decade. As a mom of four, value, usefulness and online safety take priority. Have a question? Email Leslie at asklesliemeredith@gmail.com.