Candidate Fraud Is Getting Smarter — SmartRecruiters Responds with Behavioral Detection
The News
SmartRecruiters has unveiled a prototype fraud-risk tool that uses behavioural signals, device intelligence, and network indicators to detect suspicious job applications, the company announced at its April product preview. The tool is designed to address a growing industry problem: AI-generated resumes, bot-submitted applications, and fraudulent candidate profiles that waste recruiter time and introduce security risk.
The system analyses patterns across submissions — identifying when multiple applications share device fingerprints, exhibit unusual application velocity, or contain AI-generated content matching known fraud patterns. High-risk applications can be quarantined, routed for additional verification (including an automated AI interview), or sent to a manual review queue. Configurable thresholds let teams tune the system to their risk tolerance without blocking legitimate applicants.
SmartRecruiters joins a growing number of TA platforms adding fraud detection capabilities, reflecting industry recognition that the “authenticity dilemma” — as Deloitte’s recent report terms it — is no longer hypothetical. Fake applications are real, they are increasing, and they are costing TA teams time and trust across the hiring funnel.
Why It Matters
If you have not encountered a fraudulent application in your pipeline yet, you may just not have looked closely enough. HR Dive reported earlier this year that AI enables candidates to tailor resumes to match job descriptions exactly — and that fake applicants are submitted by bots, sometimes without real people’s knowledge. As screening tools get smarter, the methods used to bypass them evolve in parallel.
For TA leaders, fraud detection is becoming table stakes. A fraudulent hire in a regulated industry can have compliance implications. Even a wasted week of screening on fake applications is a direct cost to team productivity. The SmartRecruiters announcement signals that major platforms are treating this as core functionality, not a bolt-on — and if your current ATS does not offer it, that is a procurement conversation worth having at your next contract renewal.
Taking Action
All action is situated in your corporate context. These are starting points, not prescriptions — choose what fits.
Connect
Who to talk with in your enterprise
Information Security / Cybersecurity. They see fraudulent applications as a threat vector — identity fraud, IP theft, social engineering. Ask whether they have existing frameworks or vendor relationships that could extend to hiring context.
Procurement / Vendor Management. When your ATS contract comes up for renewal, fraud detection capability should be on the requirements list. Procurement can help you understand whether vendors’ fraud features are priced separately or included in your current tier.
Sourcing Operations. They are the ones most directly affected by fake applications muddying pipeline metrics. Ask what they are seeing in the data — if fraud is on the rise, they already know, even if the numbers are not being reported upward.
Inspect
What to take a closer look at
Sample your last 100 applications to a hard-to-fill role. How many have inconsistencies you could not easily verify — mismatched employment dates, skills that do not align with work history narratives, contact information that does not resolve? If the number is higher than you would like, manual sampling is not a detection strategy. Ask your ATS vendor what fraud detection capabilities exist (even if not advertised) and test them against real-world application data.
Reflect
What to think about more deeply
Fraud detection systems will inevitably flag legitimate candidates who do not fit expected patterns — gig workers, career switchers, people applying from shared devices or public Wi-Fi. The question is not whether the tool works. It is whether your review process is designed to catch false positives as efficiently as it catches real fraud. Over-blocking can cost you good candidates faster than under-detection.