The hiring landscape changed faster than most companies realised. While talent teams were still optimising for speed and friction reduction, a different kind of candidate was quietly gaming the process — and the numbers are significant.
At Running Remote 2026, Ophir Samson, Founder, CEO and CTO of Ezra, and Ben Coleman, CEO of Reality Defender, delivered one of the most sobering, practical sessions of the conference. The topic: candidate fraud in remote hiring, and what distributed companies need to do about it now.
The scale of the problem
Ben Coleman opened with a statistic that landed hard in the room. Candidate fraud is now a top-five issue for HR heads and CISOs globally. Not a niche security concern, not a future risk — a present-day operational problem affecting organisations at every size.
When Ezra runs inbound applicants through their screening process, 50-70% drop out. Validation work has confirmed those dropouts are overwhelmingly fraudulent candidates. The financial exposure isn’t limited to a bad hire, either. The speakers pointed to brand reputation damage running into the tens or hundreds of millions, with worst-case scenarios involving a fraudulent engineer gaining escalated system access over years before triggering a catastrophic data breach. France recently lost biometric data for the entire country through exactly this kind of long-game infiltration.
What detection actually looks like
The good news: sophisticated detection doesn’t have to be the first line of defence. Making detection tools visible at the point of interview acts as a deterrent that filters out 80-90% of bad actors before any technical analysis is needed. Transparency itself is a strategy.
Reality Defender operates at the technical layer, combining deepfake detection, identity verification, and ongoing surveillance. Their infrastructure requires just two lines of code to integrate — similar in approach to Stripe or Twilio — with 50 free scans per month and additional capacity available on request. At less than a penny per scan, universal deployment across all interviews and HR conversations is financially viable today.
Ezra built their platform to understand candidates beyond resumes, using voice AI interviewing with fraud detection as an integrated outcome rather than a bolt-on. Their approach flags probabilities rather than certainties. No system can offer 100% confidence, and any vendor claiming otherwise should raise immediate questions. What Ezra surfaces are timestamps and behavioural patterns consistent with fraud, presented to human recruiters for final judgement, not automated rejection.
The governance challenge
This is where the session went deeper than most fraud discussions do. Detection is not a single team’s problem.
Getting this right requires employment law, security, TA operations, and leadership aligned on what the process looks like from flagging through to decision. One organisation described what started as 10% of one person’s time evolving into a full-time dedicated TA operations role. It is a serious ongoing function, not a one-time implementation.
Several complicating factors were raised. Detection signals are probabilistic. AI-assisted tools that raise flags might also be helping candidates with accessibility needs, which requires careful evaluation to maintain inclusion. In the absence of established case law, companies — particularly government contractors — face real legal exposure if they wrongly disqualify candidates.
Choosing the right vendor
The speakers were direct about the vendor landscape. Many products now claim deepfake detection capability without the research to back it up. Running a transcript through a general-purpose LLM and surfacing a result is not fraud detection. It is a scientific problem that requires PhD-level research and peer-reviewed publication. If a vendor cannot point to published research, treat that as a red flag.
Questions worth asking any vendor: What unsolved problem are you addressing? Why are you pursuing it? How do you make money, and does that create a vested interest in misrepresenting the scope of the problem? Test solutions yourself. Validate false positive and false negative rates before buying.
The experience is two-directional
One thing the session surfaced clearly: candidates are navigating fraud risks of their own. LinkedIn shuts down thousands of fake companies daily. Fraudulent job postings designed to harvest personal information are widespread. In this environment, candidates actively prefer companies with visible security practices in place. Transparency signals trustworthiness, not paranoia.
There is also a nuanced shift in candidate preference. When given a choice between an AI-conducted interview with a three-day response time versus a human recruiter process taking seven days, candidates increasingly choose speed. They are also using AI interviews to ask the questions they feel uncomfortable raising with a human — compensation details, PTO, benefits — treating the AI as an efficient information-gathering tool rather than a judgement-rendering one.
Remote-first companies have an advantage
Distributed companies, particularly those with experience hiring across borders, have already built zero-trust practices into their operations. The shock of remote hiring fraud hits harder in organisations that spent decades assuming physical co-presence was a fraud prevention mechanism.
What to do now
The practical starting points are clear. Be transparent about your detection practices in the hiring process — the deterrent effect alone removes most bad actors. Build a cross-functional team that owns this, not just TA. Evaluate vendors rigorously. Extend fraud detection beyond the point of hire: insider threats are part of the same challenge, and screening should apply to all workers, contingent and full-time alike.
In ten years, running interviews without deepfake detection will look as reckless as operating a business without antivirus software. The technology is already cheap, the integration is simple, and the risk of not acting is substantial. The question is not whether your organisation needs to think about this. It is how fast you are going to move.
