← All guides Under the hood

Deepfake vs AI-generated: not the same problem

One invents a person who does not exist. The other puts a real person somewhere they never were. They need different checks, and conflating them causes real mistakes.

· 9 min read · Best AI Image Detector

A generated image invents its subject. A deepfake keeps a real person's identity and changes what they appear to be doing. The second is the one that harms someone specific, and it is the harder of the two to detect because most of the frame is genuine.

Two different things people call the same thing

A fully generated image starts from nothing. There is no photograph underneath it, no sensor reading, no moment that happened. Every pixel was produced by a model, and the person in it has no identity to defame because there is no person.

A deepfake starts from something real. There is a photograph or a video underneath, a genuine capture of a genuine moment, and a model has replaced a specific region of it, usually a face, with a rendering of somebody else.

The distinction matters legally, editorially and technically. A generated stock photo of a nonexistent doctor is a licensing question. A photograph of a real doctor with their face pasted onto a body they were never attached to is a defamation question, and possibly a criminal one.

Generated stock photo AI headshot of yourself Generated fake celebrity Face-swapped photo Altered document
How much of the frame is real → ↑ Harm to a specific person
Positions are illustrative. The axes are what matter: harm tracks identity, not how synthetic the pixels are.
Where each sits. The bottom-right quadrant is where the serious harm concentrates.

Why the deepfake is harder to catch

A detector reads the statistical texture of an image and reports how much of it looks like something a model produced. When the whole frame is synthetic, that signal is everywhere, and the score is unambiguous.

A face swap inverts the problem. Ninety per cent of the picture is a real photograph, with real sensor noise and real compression history, and the whole-frame average reflects that. A convincing swap can produce a score in the thirties on an image that is genuinely deceptive.

That is not a failure of the model so much as a failure of the question. Asking whether the image is generated has the wrong shape when the honest answer is that most of it is not, and one important part of it is.

16 12 19 14 11 91 17 13 15 18 12 16

The averaging that produced 33 is exactly what hides the one tile that matters.

A real photograph with one face replaced. The whole-frame score was 33 and told you almost nothing.

The region map is the answer to that shape problem. Scoring overlapping tiles separately means a localised replacement is measured against its own neighbourhood rather than diluted across a frame it does not belong to.

The vocabulary, in the order it appeared

Terms people use, and what each one actually describes
TermWhat it meansIs the subject real?
DeepfakeA real person's likeness moved onto other footageYes
Face swapThe common form of a deepfake, on a still or videoYes
AI-generatedProduced from a prompt with no source photographNo
AI-editedA real photo with something added or removedUsually
Synthetic mediaUmbrella term covering all of the aboveDepends
CheapfakeReal media misled by caption, crop or old dateYes

The last row is worth dwelling on, because it is the most common form of visual misinformation and the one no detector addresses. A genuine, unedited photograph from a different country three years ago, captioned as today's news, will score as the real photograph it is.

What to actually do, by case

  1. You suspect the whole picture is invented. Run a whole-frame check. A high score with a flat map is about as clear as this gets.
  2. You suspect a person was inserted or swapped. Read the region map first. The number will understate it.
  3. You suspect a document was altered. Same as above, and expect exactly one hot tile over the changed field.
  4. You suspect the caption is lying. A detector cannot help. Reverse image search, then check the date and place.
  5. It is video. Different tooling entirely. Frame-by-frame still analysis misses the temporal artefacts that give real-time swaps away.

Why this distinction protects people

Treating every synthetic image as equally serious produces two failures at once. It wastes attention on harmless things, such as somebody using a generated portrait because they could not afford a photographer, and it under-weights the case that actually damages a person.

The harm is not in the pixels being artificial. It is in a real, named individual being shown doing something they did not do. That is the thing worth escalating, and it is the case where a middling score is least trustworthy as a summary.

Framing an internal policy around identity rather than around technology makes it age better too. The methods will keep changing; the question of whether a specific person is being misrepresented will not.

How a face swap is actually made

Understanding the production helps explain the detection. A swap starts with a target photograph and a set of images of the person whose face will be used. A model learns to render that face under the lighting, angle and expression already present in the target.

The rendered face is then composited back into the original frame, and the seam is the hard part. The two halves have different noise, different compression history and often slightly different colour response, so the blending step smooths the boundary until an eye stops noticing the join.

That smoothing is what a region map reads. Blending removes the high-frequency texture that a sensor produces, and the resulting patch is statistically unlike everything around it even when the join is invisible. The seam is invisible and measurable at the same time.

It also explains why a swap survives a whole-frame check so easily. The manipulated area is small, deliberately blended into its surroundings, and surrounded by an authentic photograph that dominates any average taken across the frame.

The mistake that keeps recurring

People run a suspected deepfake through a detector, see a score in the thirties, and conclude the image is genuine. That is the wrong inference from the right number: the score is correct and the question it answers was never the one being asked.

The habit that fixes it is small. On any image where the concern is about a specific person, open the region map before reading the headline figure, and treat the highest tile rather than the average as the finding worth acting on.

Questions people ask

Is every AI image a deepfake?
No, and the difference is the subject. A deepfake uses a real person's likeness without their involvement, which is what makes it harmful and often unlawful. A generated image of somebody who does not exist has no victim, and treating the two the same wastes attention on the harmless case.
Can this tool detect a deepfake?
On a still image, often yes, through the region map rather than the score. A swapped face is a localised replacement inside a genuine photograph, so it shows as one hot tile in an otherwise cool grid while the whole-frame number stays low. Read the map first on any image of a person.
What about deepfake video?
That needs different tooling. Video manipulation leaves temporal artefacts, inconsistencies between consecutive frames, that a still-image model does not look for. Checking individual frames will catch some cases and miss the ones designed to survive exactly that check.
Why did a known deepfake score low?
Because most of it is a real photograph. The whole-frame score averages across the entire image, and a face occupying an eighth of the frame moves that average very little. This is the single most common way people misread a result on a swapped image.
Is a cheapfake detectable?
Not by a detector, because there is nothing synthetic to detect. A real photograph with a false caption is a provenance problem: reverse image search to find earlier copies, then check whether the date and place match the claim being made.
Does the law treat them differently?
Increasingly, yes. Several jurisdictions have introduced offences specifically covering non-consensual synthetic depictions of real people, particularly intimate imagery, while generated images of nobody in particular are handled under ordinary rules. Ask your own legal team rather than relying on this page.