As of 13 August 2026, AI can only partly detect whether a photo is fake.
This still needs a person who signs their name to it.
Can you do it?
5 minutesto a draft.
30 minutesto something you’d act on.
Cost, all in£0
Skill neededchat-fluent
Who has to check ityou
What the alternative costsThe supplied tool list does not give a price for a specialist photo-authentication service.
If this goes wrong, you may trust a manipulated image or wrongly accuse someone of creating one.
What to actually do
Hand it to a person
The route this page recommends
A person who owns the outcome does this end to end, worth it when the failure is dear.
Use a tool built for this
Second choiceDo it yourself
The distant thirdA chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.
How to actually do it
- Save the highest-quality original file you have, rather than a screenshot or compressed copy, and note where you received it and when.
- Open an image-capable chatbot and paste the prompt, then attach the original photo and add the known source, date, caption and surrounding context.
- Ask the chatbot to distinguish possible editing, AI generation, staging and misleading context instead of treating all four as the same claim.
- Inspect the file details on your phone or computer and compare the creation date, modification date, dimensions and camera information with the claimed source history.
- Use a reverse image search to look for earlier versions, different captions or the same scene appearing elsewhere, then paste relevant findings into a second chatbot message.
- Compare the chatbot's claims against the actual pixels, metadata and source evidence, and record which clues remain unexplained before deciding whether to share, buy, report or accuse.
Prompt
Analyse the attached photo for signs that it may have been digitally altered, AI-generated, staged or taken out of context. Do not give a binary answer unless the evidence is unusually strong. Separate your response into: 1) visible clues in the image, 2) clues that could have ordinary explanations, 3) what the file metadata can and cannot establish, 4) checks I can perform using the original file and source, and 5) a confidence level with the reasons for it. Do not invent metadata, source history or facts that are not visible in the image or supplied below. Treat this as a preliminary assessment, not proof of authenticity or fraud. Context and source information: [paste what you know here]. Original file details, if available: [paste them here].
Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.
What it gets wrong
- Cannot prove that a photo is genuine from appearance alone, because convincing edits can leave no obvious visual clue.
- Cannot recover missing provenance when the original file, creator, upload history or surrounding event is unknown.
- Cannot reliably distinguish a staged real photograph from an unstaged one without evidence outside the image.
- Cannot replace specialist forensic examination when the result affects a dispute, accusation, insurance claim or legal case.
What caps this at PARTLY: verification cost, judgement under ambiguity and stakes of error.
How we scored this
Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.
| Axis | Score (0–2) |
|---|---|
| Output | 2 |
| Inputs | 2 |
| Verification | 1 |
| Liability | 1 |
| Effort delta | 1 |
| Total | 7 / 10 |
The methodology and its thresholds are published in full.
FAQ
- Can AI tell if a photo is fake?
- Partly. AI can flag visual inconsistencies and likely manipulation, but a plausible result is not proof that a photo is genuine or fake.
- How accurate is AI at detecting fake photos?
- There is no single accuracy figure that applies to every model, image and type of manipulation. Treat the result as a shortlist of clues, then check the original file, source history and earlier versions yourself.
- Can ChatGPT check whether a photo has been edited?
- It can inspect an uploaded image and describe signs that may indicate editing or synthetic generation. It cannot reliably establish the complete editing history or prove authenticity from the image alone.
- What should I do if I think a photo is fake?
- Keep the original file, record its source, inspect its metadata and search for earlier versions before drawing a conclusion. Do not accuse someone or make a consequential decision solely from an AI assessment.
Nearby answers
Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.
The newsletter
AI news, new answers and product picks, straight to your inbox.