309 companies hired people who were never real

The resumes were written by an AI that read the job description first. The headshots were generated and placed into stolen identity documents. During interviews, voice-changing software masked the accent in real time. Every one of those companies ran a hiring process. Every process was cleared.

In one U.S. Justice Department case, 309 businesses were defrauded using 68 stolen identities. Gartner projects that by 2028, one in four candidate profiles worldwide could be fake.

The pain point is not the fraud. It is what your screen measures.

Look at what was faked: a face, a voice, a resume, an interview answer. Every one of those is a stand-in for capability. Not one of them is capability.

And the fraud did not stop at the offer letter. Microsoft documented these operatives prompting AI every day after being hired to answer technical questions and write code, because the one thing they could not fabricate in advance was doing the job.

Your organisation is not a target of a state-backed operation. But if that operation can clear your screen, so can the candidate with ChatGPT open on a second monitor.

The best-sounding answer is often your worst Hire

Ask any AI how to handle an escalating client. You will get total empathy, total ownership, unlimited persistence — no trade-offs, no point of escalation.

On a 1-to-5 scale, that is nearly a perfect answer.

On the floor, it describes someone who never escalates and absorbs pressure until they stop absorbing it. That is not a strong Customer Service Executive. That is your next resignation, and a measurable Hiring Impact you will absorb for two quarters.

💡 This is why single-scale scoring fails. It cannot tell High Good strength with calibration from High Poor, the same trait at full intensity with none. PMaps scores both separately, across 73 competencies and 255 live roles, so you see which kind of high you are hiring.

A candidate can generate an answer. They cannot generate a disposition.

This is the part worth holding on to.

An interview however well designed asks for a response. Responses can be rehearsed, coached, or generated. A psychometric assessment does not ask for a response at all. It measures trait patterns across many items, scored against how people who succeed in that role actually behave.

There is no correct answer to retrieve. No job description to mirror back. AI can help a candidate write about their judgement. It cannot supply their judgement, their stress response, or their disposition.

💡 Two instruments, two jobs. 

Psychometric and cognitive assessment runs first and measures the person  traits, reasoning, domain ability, against a role benchmark.

AI Video Interview then scores behavior on the four anchors, with identity verified continuously through the session rather than once at the start.

Together they close the gap the fraud exploits: one measures what cannot be fabricated, the other verifies who is in the room and how they actually reason.

📋 How to make the improvement

• Move measurement before the interview, so the conversation stops carrying the screening load. 
• Score against a benchmark for that role at that level not a general impression. 
• Use formats with no retrievable answer: situational judgement, cognitive, behavioral. 
• Ask every vendor: can your scale represent a candidate who is strong in a way that derails? 
• Verify identity continuously through the interview, not once at the start.

Efficient screening is not faster screening. It is screening that measures the work  psychometrics for what cannot be faked, video interview for how they actually reason.

🎁 2 giveaways for YOU this month

1. The 4-Anchor Cheat Sheet — High Good, High Poor, Low Good, Low Poor, with worked examples
2. 10 free pilots, for the first 100 Clients in this round  on AI Video Interview or psychometric assessment, your choice

To claim a pilot, email ssawant@pmaps.in and tell us which of the two you want to test  or take both on the same role and compare. No cost, no investment, no commitment. One live role, your candidates, your report.

A question worth sitting with: of your last ten Hires, do you know which ones scored well the right way?