MediaChecker
An older gentleman’s arms resting beside a printed photograph of a couple walking by water

Is It Real? Or Is Something Else Going On?

MediaChecker analyzes images and video for signs of synthetic generation, manipulation, transformation, recapture, file degradation and other forensic inconsistencies. Because the question is rarely as simple as “real or fake.”

Check Your MediaHow It Works

Check Before You Believe. — A photograph can be real and still be misleading.

Check Before You Believe.

A genuine camera image can be edited. An authentic image can be reposted with false context. A screenshot can obscure the original file history. An AI-generated image can imitate the characteristics of a camera.

MediaChecker does not pretend every digital file has a simple answer. It examines the evidence.

Check Images and Videos Before You Trust Them

  • Was this image generated by AI?
  • Has this photograph been materially manipulated?
  • Is this a camera-derived image or a reconstruction?
  • Has this media been copied, re-encoded or recaptured?
  • Does the available evidence support the story being told about the file?

MediaChecker is not a “truth machine.” It is an evidence-analysis system designed to help people understand what may have happened to a piece of digital media.

15 free analyses per month · Premium analysis available

More Than an AI Detector

Most detection tools ask one question: was this made by AI? MediaChecker examines a broader problem.

  • AI-generated
  • AI-assisted
  • digitally manipulated
  • composited
  • inpainted
  • heavily re-encoded
  • screen-captured
  • photographed from another screen
  • stripped of metadata
  • reposted with misleading context

Those categories are different. A manipulated photograph is not necessarily AI-generated. An AI-assisted image is not necessarily entirely synthetic. Missing metadata does not prove deception. A low-quality screenshot is not evidence of manipulation.

The File Is Not Always the Story.

A real photograph can be paired with a false caption. An original image can be reposted without attribution. A viral image can pass through multiple platforms before you see it.

Digital verification is not just about asking whether pixels are real. It is about understanding the evidence available — and understanding what that evidence cannot tell you.

Distribution is a separate question

Eight Engines. One Evidence Framework.

Independent systems rather than a single authenticity percentage. Forensic observations, confidence, interpretation and policy stay separate so that no single signal is allowed to make the entire decision.

  • 01

    Media Forensics

    Examines measurable characteristics within digital media to identify relevant forensic evidence.

  • 02

    Forensic Coherence

    Evaluates whether independent observations are consistent, contradictory or inconclusive.

  • 03

    Evidence Aggregation

    Organizes observations from multiple analytical systems without reducing complex evidence to a single simplistic metric.

  • 04

    Policy

    Determines what the available evidence is allowed to support — and when the system must remain inconclusive.

  • 05

    Uncertainty

    Updates confidence as evidence arrives. Incomplete files stay uncertain. A probability is not a fact.

  • 06

    Behavioral Hesitation

    Looks for uncertainty in the story around the media. Hesitation is not guilt. Research architecture — human interpretation required.

  • 07

    Collective Judgment

    Considers whether agreement around a file appears independent. Consensus is not automatically treated as truth.

  • 08

    Adversarial Resistance

    Asks whether the decision-making environment itself can be trusted. We do not publish the methods.

The eight engines

Same Scene. Different Digital Histories.

Two images can depict the same subject while having completely different forensic characteristics. MediaChecker does not ask whether the subject looks familiar. It examines the digital evidence surrounding the file itself.

Original camera photograph of horses in a paddock used as a native-capture reference

Is this real? Yes.

Our analysis shows this image is most likely a real photo taken by a camera.

Real photoLow risk

What we found

87%

Real photo

  • 87%Real photo

    The image looks like it was taken by a camera, with no strong signs of AI generation.

  • 7%Possible edits

    Small changes like brightness, color, or cropping may have been made.

  • 6%Unlikely AI

    Very low signs that this image was created by AI.

Why we think this

  • Camera evidence

    The image has natural camera patterns.

  • No AI patterns

    No signs of AI-generated textures or artifacts.

  • Consistent details

    Shadows, lighting, and perspective look natural.

  • File information

    The file contains normal camera data.

Plain language verdict

Our analysis shows this image is most likely a real photo taken by a camera.

Overall verdict

Real

MediaChecker looks at many clues, not just one. Results are not 100% guaranteed. Learn more about how we analyze media.

AI-generated artistic transformation of the same paddock photograph

Is this real? Probably not.

This looks like it was made or rebuilt by AI, not photographed as you see it now. It may still show a real place or person.

Likely AICheck carefully

What we found

63%

AI-made

  • 5%Real photo

    Some camera-like detail remains, but it is not the main story in this file.

  • 32%Possible edits

    There are signs the picture was changed after it was made — cropping, retouching, or compositing.

  • 63%AI-made

    The file shows patterns common in AI-generated or AI-reconstructed pictures.

Why we think this

  • Weak camera evidence

    It does not look like a first photograph from a camera.

  • AI patterns

    Textures and structure match AI-generated pictures.

  • Rebuilt details

    Lighting and edges look constructed, not captured.

  • File information

    Little or no camera file data. That supports a generated file, but is not proof by itself.

Plain language verdict

This looks like it was made or rebuilt by AI, not photographed as you see it now. It may still show a real place or person.

Overall verdict

Likely AI

MediaChecker looks at many clues, not just one. Results are not 100% guaranteed. Learn more about how we analyze media.

What MediaChecker Can — and Cannot — Tell You

MediaChecker analyzes digital evidence. It cannot independently prove:

  • who created an image
  • where an image was taken
  • when an image was taken
  • whether a caption is true
  • whether a person is lying
  • the motive behind an image
  • the complete history of a file

A forensic analysis describes the evidence available in the media provided. It should be considered alongside source verification, context and independent reporting.

ICBM Magazine

Learn How Digital Media Actually Works.

MediaChecker analyzes the file. ICBM Magazine explains the larger problem: AI-generated images, deepfakes, manipulated media, misinformation, provenance and digital literacy.

Explore ICBM Magazine

Check Before You Believe.

Upload an image or video and examine the evidence behind it. Not a guess. Not a single AI percentage. A structured analysis of what the available evidence can — and cannot — support.

Check Your Media

15 free analyses per month.