What it is
A ghost candidate is an application built around an identity that does not correspond to any real person: an invented name, a stitched-together work history, a CV or profile generated or assembled specifically to exist, not to represent someone who actually wants the job. Some are built for volume, an agency or platform channel that needs application numbers to look healthy, so an operator submits dozens of synthetic profiles against real roles to inflate throughput. Others are financial: a ghost candidate reaches offer stage and a fee gets paid, or the profile is used to harvest data or run a scam before anyone realises there was never a start date to reach. The unifying trait is that there is nobody to place, however far the process runs.
A ghost candidate is not a deepfake candidate. A deepfake candidate is a real applicant whose voice, face or resume has been synthetically altered or impersonated using AI to get through screening; take away the fakery and a real person with a genuine interest in the job remains. A ghost candidate has no such floor: there is no real person under the fabrication at all. It is also not the same as candidate ghosting, where someone who was genuinely engaged goes silent partway through a process; a ghost candidate was never really there to begin with. Think of it as the candidate-side mirror of a ghost job: a real employer posting a role it has no intention of filling, met from the other side of the desk by an application that was never a real person applying for it.
A ghost candidate does not oversell who they are. There is no one there to oversell.
Why it matters
A ghost candidate pollutes the top of the funnel exactly where volume is easiest to mistake for progress: a shortlist that looks full, an application count that looks healthy, a submission that clears an initial CV screen can all rest on an identity that will never sit an interview honestly, or never existed to start a role at all. For an agency the exposure is direct: time spent building a profile that goes nowhere, a submission put in front of a client that turns out worthless, or worse, a synthetic candidate that clears screening and reaches offer stage, at which point the agency's name and fee sit behind a hire that was fabricated from the start.
Generative AI has lowered the cost of building a synthetic profile that reads as plausible on paper, a name, a work history assembled from real job descriptions, an email and phone number that pass basic checks, to something that can be attempted at real volume rather than as a rare one-off. That sits inside a broader, well-documented rise in AI-enabled hiring fraud generally, alongside the parallel rise of deepfake interview scams, and it means a recruiter can no longer assume that a CV which reads correctly belongs to a person who is actually there. The practical cost is trust: once a desk has been burned by one ghost candidate that made it past an initial screen, every subsequent unfamiliar application picks up a small tax of suspicion, whether it deserves it or not.
How boilr handles it
boilr does not claim to be an identity-fraud detector, and no honest product does. What it does is design against the gap a fabricated identity relies on. Candidates and contacts that reach a task in boilr are sourced through cross-referenced signals and public data across the sources boilr watches, not accepted at face value from a single self-submitted profile, so a name with no footprint anywhere else is exactly the kind of gap this stage is built to surface before a consultant spends time on it. Every contact also runs through boilr's data waterfall and is checked against multiple sources, so an identity that only exists on the one document a ghost candidate can control is treated as unverified rather than as fact.
boilr is human-in-the-loop by default: nothing reaches a client until a consultant has reviewed the task, and the Company Brain carries the agency's own placement and engagement history alongside it, so a pattern that doesn't fit, an identity with no verifiable trail, a profile that never resolves to a real, reachable person, gets flagged against what the desk has actually seen before, rather than waved through because the pipeline needed a number. That combination doesn't replace judgement. It means judgement starts from a shortlist worth spending it on.
