TrueMedia.org - Deepfake Analysis

Get the full experience

Get a free account to upload audio, video and image files and save history.

Is this Real?

An analyst reviewed this: Real.

apnews.com/article/walz-family-nebraska-trump-elelction-2f2e29586b0e9ebe094ff2cf546e1ae4

Verifiable claim in the image

From X · original post (opens in new tab)

AnalyzedWed, Sep 4, 2024

Detector time1m 30s

File.jpg143KB

Disclaimer: truemedia.georgetown.edu uses state-of-the-art academic AI methods. However, errors can occur.

TrueMedia

Media Analysis Report · An open-source research project of Georgetown University

J4_EzcHg_r5objzaiIJPMaM1hOk.jpg

Wed, Sep 4, 2024

Preliminary — analyses are still in progress; the verdict and signal below may change

Low Manipulation Signal

LowManipulation possibility

File: .jpg · 143KB
Scope: Image
Source: https://twitter.com/JamesZClarkEsq/status/1829971229966545192
Analyzed: Wed, Sep 4, 2024 · 1m 30s
Detector agreement: None completed

Detector	Model	Result

Disclaimer: truemedia.georgetown.edu uses state-of-the-art academic AI methods. However, errors can occur.

TrueMedia · Media Analysis Report — Interpretation

J4_EzcHg_r5objzaiIJPMaM1hOk.jpg

Verdict tiers

Little Evidence	Low possibility of manipulation
Conflicting Evidence	Moderate possibility of manipulation
Substantial Evidence	High possibility of manipulation

Manipulation possibility scale

Low

0–20

Noticeable

20–33

Moderate

33–67

Strong

67–85

Nearly certain

85–100

This analysis: Low manipulation possibility. This reports how strongly the detectors flagged this media — it is not a probability that the media is manipulated. The verdict itself is determined by TrueMedia's vote-based aggregation of detector results.

Methodology

Each detector listed on page 1 independently analyzes the media for signs of AI generation or manipulation. TrueMedia aggregates the detector results into the overall verdict using a vote-based scheme in which trusted detectors carry extra weight; the manipulation possibility shown is a calibrated summary of the detector scores, reported as a strength rather than a probability. In this analysis, 0 of 0 completed detectors agreed with the final verdict.

Cite this analysis

@misc{truemedia2026,
  title        = {TrueMedia: Open-Source Media Manipulation Detection},
  author       = {{Georgetown University}},
  year         = {2026},
  howpublished = {\url{https://truemedia.georgetown.edu}},
  note         = {Analysis J4_EzcHg_r5objzaiIJPMaM1hOk.jpg}
}