Verification built for newsrooms.
HMNPROOF is a free, nonpartisan authenticity scanner for journalists, fact-checkers, and editors. Paste a URL or upload a file — get a probabilistic risk report with a signal-by-signal breakdown, an independent ensemble cross-check, and a public verification link you can cite in your reporting.
Nonpartisan by construction
Verdicts are independent of subject, platform, or viewpoint. The same image returns the same score regardless of who is depicted.
Ensemble, not one model
Two independent vision models score each scan with blind prompts. Agreement, partial agreement, and disagreement are surfaced honestly.
Citable public reports
Every verdict gets a stable /r/{id} URL and a branded PDF — something you can link in a story, hand to an editor, or attach to a correction.
Web-match evidence
SerpAPI-backed reverse-image lookup shows where else this media lives online, with thumbnails, source links, and earliest known appearance.
Provenance-first
C2PA Content Credentials, EXIF/XMP, byte-level forensics, and AI-generator fingerprints are read directly from the file bytes.
Disclosed limitations
Every report lists what the technology cannot detect. We never claim a verdict is forensic certification — only probabilistic guidance.
What HMNPROOF is built to support
Breaking-news verification
Get a fast probabilistic read on a viral image or video before it ends up in your lede. Includes reverse-image search and source-host reputation.
Pre-publication fact-checking
Run any AP, Reuters, or wire image through the ensemble to confirm authenticity signals before running with it on a politically sensitive story.
Correction support
When a story needs to be amended, the public /r/{id} link gives editors and readers an auditable third-party record of the analysis.
Source verification
Confirm a tipster's photo or video matches its claimed origin device, claimed timestamp, and claimed first appearance online.
Election & political-imagery checks
Detect AI-generator provenance markers, double-compression, and synthetic faces across political content with a nonpartisan scoring model.
Newsroom training
Use the methodology page and ensemble disagreement reports as teaching material for reporter training on AI-era verification.
Press kit — at a glance
- What it is
- A free, nonpartisan authenticity risk scanner for images, video, URLs, text, and audio.
- Who it's for
- Journalists, fact-checkers, deepfake victims, creators, and the general public.
- What's free
- Unlimited scans, public verification links, ensemble cross-check (signed in), and PDF reports for signed-in users.
- Detection stack
- C2PA, EXIF, byte-level forensics, two independent vision models, reverse-image web match, source-host reputation, Bayesian learning engine.
- What we don't do
- We do not train, sell, or license generative AI models. No conflict between selling cloning and selling detection.
- Editorial stance
- Probabilistic guidance, never declarative. Limitations are disclosed on every report.
- Headquarters
- Independent. United States.
- Contact
- press@hmnproof.com
Quoting HMNPROOF
"HMNPROOF returned a {score}% AI-generation risk score for the image, with {agreement}% agreement between two independent vision models. The full report is available at hmnproof.com/r/{id}."
When citing a HMNPROOF verdict, please include the report URL so readers can see the full signal-by-signal breakdown. Verdicts are probabilistic; we do not provide legal or forensic certification.
Media inquiries & expert commentary
For interviews, expert commentary on AI media, on-deadline verification help, or partnership inquiries from newsrooms and fact-checking organizations:
We aim to respond to deadline inquiries within two hours during US business hours.