Attack Vectors

What and how generative AI techniques are used to deceive, impersonate and defraud.
Back to Attack Vectors

AI image manipulation and inpainting

Image
Corporate Fraud
Consumer Fraud

AI image manipulation uses generative models to add, remove, or alter elements of a real photograph. Inpainting tools fill selected regions with AI-generated content that blends seamlessly into the surrounding image.

Content advisory: this technique is used to produce AI-generated child sexual abuse material (CSAM), including sexual imagery of real children and fully synthetic depictions. Creating, possessing, or sharing this material is illegal. Report suspected CSAM to the NCMEC CyberTipline at report.cybertip.org.

Use cases: Dispute & claim verification · Media authentication & moderation

How it works

  • The user selects a region of an existing image and provides a text prompt describing the change
  • A diffusion model generates new content for the selected region while preserving the rest
  • The output looks like a natural photograph, with no obvious edit boundaries to the human eye

Where it shows up

  • Insurance claim fraud where damage is added to or removed from photos
  • Manipulated proof-of-delivery and condition photos in marketplace disputes
  • Fabricated visual evidence in legal cases, with people, objects, or text inserted or erased
  • Doctored news photos used to amplify or fabricate events
  • Modified ID documents and selfies for KYC fraud