The AI Arms Race: Navigating Hiring Authenticity with Intelligent Fraud Detection

AI has fundamentally reshaped how candidates prepare for job opportunities. What started as using generative AI to enhance resumes has evolved into candidates utilizing these tools during live interviews as scripts, creating new challenges for talent acquisition teams.

In a recent episode of Talent Experience Live, host Devin Foster explored this emerging reality with Sebastian Niewöhner, Senior Director of Product Management at Phenom, to understand how fraud detection technology can help organizations identify potential integrity issues without compromising fairness in hiring. The scope of the challenge has become significant. "Three out of four interviewers have detected suspicious behavior in the last six months," Niewöhner revealed.

This shift raises important questions about authenticity in hiring. The conversation reveals why maintaining genuine candidate assessments matters and how organizations can address these challenges while preserving fairness and humanity in their processes.

Catch the full episode here or jump into the highlights below.

The Evolving Use of AI in Job Applications

The rise of generative AI tools has created new possibilities for candidates to enhance their job search process. What began as resume optimization has evolved into more sophisticated applications during the interview process itself. "We've seen the TikTok videos where there's an iPhone in front of the camera, and the candidate is reading off a script," Niewöhner explained. "There was another example from a coding interview where the entire code challenge was done by AI."

The trend extends beyond individual cases. The competitive job market and volume of applications have created pressure where candidates feel they need to use available tools to remain competitive. As Niewöhner noted, "it's kind of even that internal discussion around, do I have to do it as well? A lot of candidates are using it to prepare, and some are also using it to kind of get an unfair advantage in an interview."

Remote interviewing has made AI assistance more accessible during the interview process. This creates challenges for hiring teams who need to assess authentic candidate capabilities while recognizing that some level of preparation enhancement is natural and expected.

This shift from authentic self-representation to scripted performance changes the fundamental nature of candidate evaluation.

3 Pillars of Modern Fraud Detection

Phenom's approach to fraud detection focuses on three core areas that work together to identify potential integrity issues without making assumptions about candidate intent. The three pillars include:

1. Identity Validation

The most concerning issue involves candidates who aren't who they claim to be. "We recently had a customer tell us about an interview where the candidate who showed up on the first day wasn't the one who participated in the interview process," Niewöhner shared.

This type of misrepresentation involves voice and video comparison across all interview stages to verify that the same person participates throughout the process. With many organizations using video interviews for screening and face-to-face meetings for final rounds, maintaining identity consistency becomes challenging without proper verification.

2. Authenticity Assessment

Rather than trying to detect AI usage directly, the system looks for patterns that suggest inauthentic responses. These include unusually long delays before answering, overuse of buzzwords, and responses that remain at a high level without specific examples.

"This doesn't necessarily mean they're using GPT — maybe it's just a bad response or they take time to think," Niewöhner clarified. "But it's an indication that something might be off, especially when we look at patterns across multiple interviews."

The technology recognizes that determining intent can be complex. As Niewöhner explained, "customers use our video interview solution for asynchronous interviews, and all the face-to-face interviews are also recorded. That's the first component: identity validation."

3. Consistency Analysis

The most sophisticated aspect involves comparing candidate responses across different interview rounds. A sales executive might claim significant success in their first interview but provide reduced numbers when asked similar questions later in the process. Through Phenom Interview Intelligence, organizations can record candidate interviews and generate assessments for team members who missed the live session.

This approach recognizes that authentic candidates tell consistent stories about their experience, while those relying on AI assistance may struggle to maintain consistency across multiple touchpoints.

Building Ethics Into Detection Technology

Creating technology that evaluates candidate authenticity requires careful consideration of bias and fairness. The challenge becomes particularly complex when dealing with candidates for whom English is a second language or those who naturally take time to formulate thoughtful responses.

Niewöhner explained their approach: "First, we have very strong internal measurements to make sure there's no signal waste if it's not above a certain threshold of confidence. These signals are communicated as potential warnings with clear explanations of why something was detected."

The system provides hiring teams with video snippets and a detailed reasoning for each signal, allowing human reviewers to make informed decisions. This approach keeps humans in control while providing them with insights they might otherwise miss. As Niewöhner noted, "it's hard sometimes to detect in a single interview if someone is cheating or not. It might just be a bad response."

Perhaps most importantly, if a hiring manager reviews a signal and determines it's incorrect, they can reject it entirely. The feedback immediately improves the system, and the signal disappears from the candidate's profile permanently.

"We're not detecting this candidate is cheating — we're giving the hiring team more data to do manual validation," Niewöhner emphasized.

Industry Response and Early Results

The technology has found particular traction in two distinct markets. Financial services companies — including banks, fintech firms, and insurance organizations — show strong interest due to their risk-aware culture and the high cost of bad hires.

"A wrong hire costs $50,000 to $70,000 down the river," Niewöhner noted. "These organizations understand that preventing fraud saves significant resources later."

The second major market involves high-volume recruiting in regions like the Middle East and India, where identity switching during interviews has become a persistent challenge. Customer service and support roles face particular issues with candidates sending friends or family members to interviews.

"The initial feedback has been great from the customer that we wrote this out to," Niewöhner shared. “What we see from the very first weeks is that 6% of the applications receive signals in the process. Also, over 60-70% of those are actively confirmed by the recruiters who are reviewing those signals. We get very good feedback that the accuracy is high.”

The Future of Authentic Hiring

Fraud detection represents just one piece of a larger transformation in talent acquisition. As AI becomes more sophisticated, the hiring process must evolve to maintain authenticity while embracing helpful technology.

The solution isn't to eliminate AI from hiring — it's to create transparent boundaries around its appropriate use. Many organizations now include guidance on their career sites about how candidates can ethically use AI in their application process.

"Being upfront that we might check for certain signals gives candidates confidence and transparency about what's happening," Niewöhner said. "This will probably get a lot of candidates to stop using inappropriate AI assistance in the first place."

The technology will continue evolving. Future developments include real-time insights during interviews, enhanced application analysis, and extending integrity checks into the onboarding process to verify that new hires can deliver on the capabilities demonstrated during interviews.

Building Trust Through Technology

The goal isn't to identify or penalize candidates who use AI assistance — it's to preserve the authenticity that makes hiring decisions meaningful. When candidates misrepresent their actual capabilities, organizations make hiring decisions based on incomplete information, which ultimately doesn't serve anyone well. "You don't want to start a relationship with an organization on the wrong foot," Foster observed during the conversation. "This is a commitment you're making, and it should be genuine."

Fraud detection technology serves as a tool to maintain that authenticity. By providing hiring teams with better information about candidate responses, it helps ensure fair evaluation while protecting the integrity of the hiring process.

The future of hiring depends on balancing technological advancement with human judgment. Fraud detection represents a step toward that balance, using AI to preserve authentic assessment rather than replace human decision-making.

As the hiring landscape continues evolving, organizations that embrace both technological tools and ethical practices will build stronger teams and more authentic relationships with their talent. The technology exists to support this vision — the question is whether organizations will choose to use it thoughtfully.

Download our Phenom Interview Intelligence Guide to learn how consistent, compliant interviews with bias-free evaluation can help your team make better hiring decisions.

Devi B

Devi is a content marketing writer who is passionate about crafting insightful content that informs and engages. When not writing, she enjoys watching films and listening to NFAK.