How can I tell if a photo is AI-generated?

AI can now create photographs of people and events that never happened, convincingly enough to fool a quick glance. They turn up in fake news, scam adverts, and hoaxes. You will not catch every one by eye, so the real skill is the same as for any surprising claim: check where it came from before you believe it or share it.

Clues in the image

Hands and fine detail. Odd fingers, teeth, ears, and jewellery are still common giveaways.
Text that is nonsense. Signs, labels, and logos in the background often come out garbled.
Backgrounds that do not add up. Warped lines, repeating patterns, or objects that merge into each other.
Skin and light that look a touch too smooth, glossy, or "perfect".

These help, but AI is improving quickly, so treat them as hints, not proof.

Check the source, not just the pixels

Where did it first appear? A dramatic image with no reputable source behind it deserves suspicion.
Do a reverse image search to see where else it has been used, and whether fact-checkers have looked at it.
Be extra wary if the image is doing a job - stirring outrage, backing a "too good to be true" offer, or pushing you to act.

If you only do one thing

Do not believe or share a striking image on sight. Check whether a reputable source is carrying it - the source is a better test than any visual clue.