GenVeritas tells you whether an image, video, voice or document is real, edited, or AI-generated — and shows you exactly why, with a signed verdict you can stand behind.
2 regions flagged · no camera provenance · metadata stripped
Anyone can now generate a convincing photo, voice or document in seconds. The cost of faking dropped to zero. The cost of being fooled didn't.
to tell a real photo from a good fake with the naked eye. You can't — and neither can your team.
one convincing fake is all it takes to move money, sink a reputation, or settle a dispute the wrong way.
new generators every month. Detection has to keep learning as fast as the fakes do.
We don't rely on a single model guessing. GenVeritas cross-checks provenance, forensics and context, then explains where the evidence points.
We read content credentials, camera signatures, EXIF and edit history to establish where a file came from — and where the chain breaks.
An ensemble of detectors looks for generator fingerprints, compression anomalies, lighting and lip-sync mismatches, and spectral traces in audio.
You get a clear verdict, a confidence score, highlighted evidence, and a cryptographically signed report you can attach to a case or audit.
Drag in an image, clip, voice note or PDF — or pick a sample. This preview runs in your browser and never uploads anything.
Get full access →Use it in the browser for one-off checks, or wire it into your workflow so every file is verified before it reaches a human.
Images, video, audio and documents run through the same pipeline and come back in the same report format.
Drop verification into onboarding, claims, moderation or newsroom tools.
Heatmaps and plain-language reasons for every flag, so reviewers can agree or push back.
Every verdict is timestamped and signed, so it can be checked later and can't be quietly altered.
Files are processed and discarded by default. Never used to train models. Regional processing on request.
If a fake file could cost you money, trust or a wrong decision, GenVeritas belongs in the workflow.
Check user-submitted photos and footage before they go to air or print.
→ 02Catch generated damage photos, edited receipts and forged invoices at intake.
→ 03Verify listing photos and host or guest IDs before a booking goes through.
→ 04Stop synthetic selfies, deepfake video calls and cloned-voice payment requests.
→ 05Assess digital evidence and screen applications with a defensible, signed record.
→Simple plans, no long contracts. Early-access members lock in launch pricing.
No — and anyone who says otherwise is selling something. That's why GenVeritas gives a confidence score and shows its evidence instead of a bare yes/no, so a human can make the final call with full context.
Re-saving and screenshots strip metadata and blur some forensic traces. We flag when provenance is missing and lean more heavily on content-level signals, and we tell you when the evidence is thin.
By default, files are analysed and deleted. We never use customer files to train models. Enterprise plans can choose where processing happens.
An open standard for "content credentials" — a tamper-evident record of how a file was made and edited. When a file carries them, we read and validate them as part of the provenance check.
We're onboarding early-access users in waves. Join the list below and you'll get access, plus launch pricing locked in.
Join the early-access list. We'll reach out when your spot opens.