AI is profoundly changing the way businesses recruit, to the point that even the concept of "candidate" is questioned.
A seemingly perfect interview with an impressive CV, smooth answers, but now it can hide a worrying truth that the applicant may not exist at all.
This warning was issued in Experian's 2026 fraud report (a global information and credit reporting service company based in Dublin, Ireland), showing that AI not only supports job search but also creates virtual candidates.
From personalized CVs to real-time deepfake video interviews, the entire recruitment process is becoming the target of sophisticated forms of fraud.
Recruitment and security problems
According to Anirban Mukerji, CEO of miniOrange (a security solution provider focused on Identity Management and Access), traditional interviews have become the most exploitable weakness in corporate security systems.
He cited Gartner's forecast (a global information technology research and consulting company based in Stamford, USA) that by 2028, about 25% of candidates could be products of generative AI.
Mr. Mukerji warned that if the recruitment process is still based only on PDF CVs and webcam interviews, businesses may unintentionally open the door to cyber attacks, causing significant damage.
When "perfect candidate" is suspicious again
Mr. Vaibhav Koul from Protiviti (a global consulting company specializing in providing internal audit services, risk management, technology, finance, headquartered in the US) believes that "too perfect" records need to be carefully considered.
Fake candidates are often built by AI with overly polished images, experiences and career stories.
Conversely, real records often have natural imperfections such as seamless timelines, diverse experiences, and clear activity traces in the digital environment.
Technical signs such as lip lag, abnormal light, or blurred image borders in the video can also be signs of deepfake.
What do businesses need to do?
Experts recommend that businesses need to apply multiple layers of identity verification. This includes verifying official papers through digital platforms, real-time biometrics, and comparing digital traces of candidates.
In addition, requiring random operations in interviews such as changing camera angles, presenting papers directly, or solving problems immediately, can help detect fraud.
The "Zero-Trust" recruitment model is also proposed, in which all information must be independently verified, instead of default trust to avoid fraud and risk.
