Visual fraud in business onboarding: what changes for verification teams?

TL;DR: AI-driven deepfakes and synthetic media are eroding trust in business onboarding, vendor onboarding and third-party risk workflows, according to Truepic, while authenticated image capture added to D-U-N-S processing helped flag suspicious applications, reach more than 87 countries and deter nearly 70% of flagged applicants from continuing. Visual authenticity is now a governance control, not a nice-to-have assurance layer.

Editorial analysis by NHI Mgmt Group, based on content published by Truepic: “Dun & Bradstreet and Truepic: Securing business identity at scale with visual risk intelligence”.

By the numbers:

• Truepic says nearly 70% of applicants asked to provide authenticated imagery either did not respond or abandoned the process.
• Truepic says 57.8% of virtual inspections contained images captured more than 200 meters from the expected address.

Key questions

Q: What breaks when business verification trusts uploaded images too much?

A: Onboarding breaks when the system validates the appearance of evidence instead of the provenance of evidence.

Q: Why do deepfakes change identity verification requirements?

A: Deepfakes change requirements because they let attackers create believable but false evidence that can pass weak visual checks.

Q: How should teams judge whether authenticated capture is working?

A: Look for whether suspicious submissions are being stopped before approval, whether applicants abandon the workflow when challenged, and whether location or image anomalies are being detected consistently.

Practitioner guidance

• Embed capture-time authenticity checks Require the verification workflow to confirm when, where and how media was captured before the submission can proceed.
• Add device and provenance attestation Bind image and video submissions to trusted capture conditions, including timestamp, geolocation and device orientation, so spoofed evidence is easier to reject.
• Layer fraud detection across the media chain Use reverse image search, anomaly detection and inspection of metadata and encryption signals together, because one control will not catch every manipulation path.

Bottom line: AI-generated and manipulated media now threaten the trust assumptions behind business onboarding, vendor onboarding and third-party risk workflows.

Explore further

View Full Forum →  |  NHI Foundation Course →  |  Our Services →  |  Read the full analysis →