isGenAI
Check for AIUpgrade

Fraud and trust teams

AI fraud detection for photos, IDs, receipts and documents

isGenAI checks the files people submit as proof (ID photos, selfies, receipts, damage photos, screenshots and PDFs) for evidence that AI made or edited them: C2PA Content Credentials, IPTC AI labels, saved prompts and generator metadata. Free, through a web checker or an API with no key.

Check a file nowRead the API docs

Fraud checks by use case

  • KYC and onboarding. ID document photos and selfies uploaded during account opening. KYC AI detection
  • Fake receipts and invoices. Expense reports, refund requests and warranty claims backed by a receipt image. Fake receipt checker
  • Fake photos. Damage photos in insurance claims, product photos in listings and profile pictures in scams. AI image detector
  • Deepfakes. Suspected deepfake images and video frames, traced back to the first upload. Deepfake check
  • Documents and PDFs. Signed Content Credentials inside PDFs, images, audio and video. Content Credentials checker
  • Video. Content Credentials and container metadata in MP4 and MOV files, such as video selfies. AI video detector

What counts as evidence

  • Content Credentials (C2PA). Signed records naming the generator and an AI source type. OpenAI says images generated with ChatGPT, Codex or the OpenAI API include them. isGenAI reports signature validity and signer trust separately.
  • IPTC digital source type. Labels such as trainedAlgorithmicMedia written by AI tools and editors.
  • Saved prompts and settings. Prompts, seeds and model names stored by AUTOMATIC1111, ComfyUI, InvokeAI and NovelAI.
  • Generator metadata and known copies. Software fields naming an AI tool, and matches with files already checked, which helps catch the same image reused across claims.

Add it to a review queue

curl --fail-with-body --max-time 30 https://isgenai.com/v1/check \
  -F "file=@claim-photo.jpg" \
  -F "privacy=private" -F "public_receipt=false"

The JSON response lists every finding with its source. Send files with AI evidence to manual review. AI agents can call the same checks through the MCP server.

Limits fraud teams should plan for

Most forged files carry no record: screenshots, re-exports and messaging apps strip metadata, and careful fraudsters remove it. isGenAI does not read SynthID watermarks, gives no visual forgery verdict and publishes no accuracy percentage. Use a finding as a reason to escalate, and keep identity, liveness and transaction checks with your existing providers.

AI fraud detection questions

What kinds of AI fraud can isGenAI help detect?

Any fraud that relies on an image, video or document file: AI-generated ID photos and selfies in onboarding, fake receipts in expense and refund claims, AI-made damage photos in insurance claims, fake screenshots and AI images in marketplace listings or romance scams. It flags files that record AI generation or editing.

How does isGenAI detect AI-generated documents and photos?

It reads evidence stored in the file: C2PA Content Credentials from AI generators, IPTC digital source types such as trainedAlgorithmicMedia, prompts saved by Stable Diffusion tools, generator and editing metadata, and matches with known copies. Each finding names its source field.

Is it a replacement for a fraud detection platform?

No. isGenAI does not score transactions, verify identities, run liveness checks or validate document security features. It adds one signal: whether the file itself records that AI made or changed it.

Why do some fake images pass with no finding?

Screenshots, re-exports and messaging apps remove metadata, and fraudsters strip it on purpose. A clean result means no record was found. It is not proof that a photo, document or person is real.

How do I add it to a fraud workflow?

POST each uploaded file to https://isgenai.com/v1/check with privacy=private and public_receipt=false. No API key or account is needed. Route files with AI evidence to manual review, and respect HTTP 429 and Retry-After because capacity is shared.