Synopsis: In this Techstrong AI video interview, Ofer Friedman, chief business development officer for AU10TIX, explains why deepfakes created using artificial intelligence (AI) technologies are becoming more difficult to detect now that professional criminals are building them.

In this Techstrong AI interview, AU10TIX chief business development officer Ofer Friedman speaks with Mike Vizard about the alarming rise of deepfake technology infiltrating the job market and corporate environments. Friedman explains that while the public hears only about the deepfakes that get caught, many more go undetected. He details how professional attackers are now leveraging deepfake-as-a-service platforms to impersonate individuals at scale, using sophisticated combinations of video, voice, and generative AI tools to bypass standard verification systems—particularly in remote hiring environments and identity onboarding processes.

Friedman emphasizes that this is no longer just amateur fraud but a systemic and organized threat, with attackers even targeting companies where identity verification was thought to be airtight. The evolving nature of these attacks makes traditional detection methods, like AI models trained on past data, quickly obsolete. Instead, he advocates for new detection strategies that identify the fingerprint of the engine creating the deepfakes, much like antivirus tools shifted to anomaly detection. He warns that even digital ID systems—thought to be more secure—are not immune, as fraudsters increasingly target verified identities and exploit weaknesses beyond the initial onboarding stage.

As the conversation concludes, Friedman urges HR departments and cybersecurity teams to collaborate and adopt multi-layered defense systems. He notes that many companies remain unaware of the threat’s scale and sophistication, even as deepfakes begin operating like long-term moles within organizations. The future of identity verification, he stresses, will depend on advanced detection methodologies, market education, and a more aggressive approach to rooting out synthetic identities already embedded in systems.