Home Blog How to Tell If a Photo Is Real

How to Tell If a Photo Is Real

Updated 2026-07-30 · 5 min read · by the FakeSec team

“Real” hides three separate questions, and mixing them up is why people get stuck. Was it photographed? Was it altered? Does it show what it claims? Each needs a different check, and only the first two live in the file.

Separate the three questions

Was it photographed at all? A generated image was never in front of a lens. This is what generation detection answers.

Was it altered afterwards? A real photograph with a region removed, added or reshaped. This is manipulation analysis — a different question with different evidence, covered in what photoshopped means.

Does it show what it claims? A genuine, unaltered photo from another place or year. No technical check catches this; only finding earlier appearances does.

Most arguments about whether a photo is “real” are actually arguments about the third question while people run tests for the first two.

The five-minute sequence

1. Reverse image search. Catches the third case immediately. Use Yandex and TinEye alongside Google — they index differently.

2. Metadata. Camera present? Dates coherent? Software field naming an editor or a generator? Free, no account.

3. Forensic analysis. Twelve methods covering generation and manipulation, with every score shown separately.

4. Look at it. Shadows agreeing, reflections corresponding, text readable, ears and teeth sensible.

5. Source. Who posted it first, and do they have any history?

Reading the result honestly

A clean forensic result means: probably photographed, probably not locally edited. It does not mean the caption is true, that the person sending it took it, or that it was taken recently.

An inconclusive result means the methods disagreed — treat it as no answer, not as clearance. Compressed screenshots land here often.

A strong AI verdict is good evidence, and still not proof. Look at how many methods agreed. Nine at 85% beats one at 99%.

When to stop checking

If the image came from a stranger, involves money, or is being used to pressure you into something, technical verification is secondary. The behaviour is the signal. People do not usually need forensic analysis to confirm what a request for gift cards already told them.

For profile photos specifically, the catfish check plus a reverse search covers both the generated and the stolen case in about a minute.

Key points

  • «Real» is three questions: photographed, unaltered, correctly described.
  • Reverse search first — it catches the case no forensic tool can.
  • A clean result means probably photographed, not that the story is true.
  • Inconclusive means no answer, not clearance.
  • When money or pressure is involved, behaviour matters more than the file.

Frequently asked questions

What is the quickest check?
Reverse image search. Seconds, free, and it catches the most common form of visual misinformation.
Can a real photo be flagged as AI?
It happens — heavily retouched or compressed photos can trigger false positives. That is why the per-method breakdown matters more than the score.
Does a camera name in the metadata prove it is real?
It supports the case but proves nothing: every field is editable with free tools.
What if all the checks disagree?
Then the image is genuinely ambiguous and you should say so rather than pick a side. Fall back on provenance — where the file came from and whether the original exists.

Check an image yourself

Reading about it only goes so far. Our AI image detector runs twelve independent methods and shows you every score, not just a verdict.

More from the blog

Try it on your own image

Free account, 3 checks included, no card required.