ooligo

Pindrop

deepfake-detection candidate-fraud-detection · interview-integrity · voice-authentication · contact-center-fraud
AI-NATIVE
Recruiting & TA
7.2 /10

What it is

Pindrop is an Atlanta voice-security company that has spent more than 15 years detecting fraud on phone calls, and it now sells the same detection for video meetings. The recruiting product is Pindrop Pulse for Meetings: an app inside Zoom, Microsoft Teams, and Cisco Webex that checks three things while an interview is running. Is this a real human (audio and video liveness, synthetic voice, face swaps, avatars)? Is it the right human (a passive match of the speaker’s voice against the voice enrolled earlier)? Is the person somewhere sensible (high-risk location and VPN signals, against a location policy you set)? It reached general availability on all three platforms on November 19, 2025, the same week TIME listed it among its Best Inventions of 2025.

The rest of the company, and most of its history, is a contact-centre fraud suite. Pulse for Contact Centers detects synthetic speech on phone calls. Protect scores fraud risk on live calls. Passport authenticates known callers by voice. BotStopper, launched September 16, 2026, tells a human caller from an AI voice agent. Pindrop says 14 of the top 20 banks use it, and it integrates with Amazon Connect, Genesys, Five9, NICE, Cisco, and Verizon. Pindrop is private. It took $100M in debt financing from Hercules Capital in July 2024, after more than $200M in equity.

Why it shows up in recruiting stacks now

  • It caught fraud in its own hiring and published the numbers. Pindrop ran its detection on its own remote engineering pipeline. It reports that 1 in 6 applicants showed clear signs of fraud and that 1 in 343 was linked to DPRK infrastructure or behaviour. Of those DPRK-linked applicants, 1 in 4 had used deepfake technology in a live interview. Read the method before you quote it: this is one company’s own pipeline, mostly engineering roles, with no sample size or period published. Its follow-up analysis concludes that AI-generated media is only a small share of impersonation. Most fraud is a proxy interviewer or a borrowed identity, not a synthetic face.
  • Its September 2026 release closed two gaps that matter for interviews. Pulse for Meetings now keeps analysing video while a candidate shares their screen, which is when coding-round proxies take over. Results also write back to the candidate record in your hiring platform. Pindrop does not name which ATSs it supports, so ask.
  • AWS sends Amazon Connect customers to it. Amazon Connect Voice ID reached end of support on May 20, 2026, and AWS names Pindrop as the replacement (see Amazon Connect). If your company’s contact centre already runs Pindrop, interview coverage can sit on an existing vendor relationship rather than a new one.
  • A court ruling reduced one biometric-privacy risk, for Pindrop. On May 12, 2026 the Third Circuit held in McGoveran v. Amazon Web Services that Pindrop qualifies as a financial institution under the Illinois BIPA exemption, because its voice authentication supports banks’ financial transactions. That reasoning is tied to banking customers. It does not obviously extend to your interviews, and it does not cover you as the employer.

Pricing reality

Pindrop publishes no prices for any product. The Pulse for Meetings page offers only a demo, and the Teams and Webex marketplace listings say a separate Pindrop Pulse licence is required. On Webex you also need a paid Webex licence. We found no free trial and no published billing unit for meetings.

What is published is the suite’s packaging. Existing Pulse for Contact Centers customers get BotStopper at no extra cost. The Pulse Warranty, with up to $1M in fraud reimbursement, requires a 3-year subscription to the full suite. That tells you how Pindrop sells: multi-year, multi-product, priced for a bank’s contact centre. A meetings-only deal for a TA team is a small contract by Pindrop’s standards, so expect little room on price and negotiate term length instead.

For scale against vendors that do publish prices: Persona starts at $250/month for 500 verifications, and Checkr adds identity verification to a $29.99 Basic report for $4.99. Our verdict: buying Pindrop for interviews alone is hard to justify. It pays off when the contact centre already runs Pindrop and the interview product is added to that contract.

Best for

A TA or talent-security lead at an enterprise, often a bank, insurer, or large contact-centre operator, whose company already licenses Pindrop for calls. They hire remote engineers, IT staff, or agents into roles with system access, run interviews on Zoom, Teams, or Webex, and treat phone screens and voice as a real exposure. Voice matching across rounds is the feature that sets Pindrop apart. It is how you catch the case where one person passes the phone screen and a different person shows up to the technical interview.

Skip it if your interviews run on Google Meet (Pindrop’s FAQ lists it as “coming soon”), if you make fewer than about 50 remote hires a year, or if you have no security or fraud team to triage flags.

Watch-outs

  • The accuracy numbers are for audio, and mostly Pindrop’s own. Pindrop claims a 99% detection rate on known deepfake engines and under 1% false positives, on its own 20-million-sample dataset. In a June 2026 Podonos benchmark it was one of four systems above 95% accuracy. No public figure exists for video or face-swap detection. And at the low fraud rates most employers see, even a 1% false-positive rate means many flags land on real candidates (the base-rate arithmetic is in the interview-fraud screening skill). Guard: pilot on your own recorded interviews with known-good and known-bad cases, and write into the hiring policy that a flag triggers an identity re-check, never a rejection.
  • Voice enrollment is a biometric, and VPN flags hit legitimate remote workers. Voiceprints are biometric identifiers under Illinois BIPA, and Illinois roles also fall under the AI Video Interview Act. The Pindrop ruling does not shift your duties as the employer. Location rules can also penalise candidates who are required to use a VPN. Guard: put notice and a written release in the interview invite, set retention and deletion terms in the contract, and configure location policy so a VPN alone never escalates. Pindrop’s own hiring team works the same way.
  • Coverage stops at three platforms plus the phone. Async video interviews and Google Meet are not covered today. Guard: map where every interview round actually happens, and make coverage of each round a written exit criterion for the pilot.
  • The suite pull is strong. A meetings pilot can turn into a pitch for a 3-year, full-suite deal. Guard: sign the meetings product as a separate SKU with a 12-month term, and let the contact-centre team negotiate the suite on its own timeline.

Alternatives, and when to pick them instead

  • An identity gate first: Persona, or CLEAR inside Greenhouse. Pick this when you can fund only one control. A document-and-liveness check at a named stage proves who the person is. Detection only flags inconsistency. See AI interview fraud for why the gate comes first.
  • GetReal Security. The closest rival on the same three platforms, with continuous face-and-voice identity verification and do-not-hire threat intelligence. Pick it when the CISO is the buyer and help-desk and executive impersonation are in scope. Pick Pindrop when voice is the main exposure or the contact centre already runs it.
  • Reality Defender. A newer detection vendor and the only one here with self-serve API tiers, from $399/month billed annually for 1,000 scans. Pick it to test detection on recorded clips before any sales cycle, or to build detection into your own interview tooling.
  • Checkr identity verification at offer. Pick it for hourly or in-person hiring where the interview is not the attack surface and $4.99 a check covers the risk.

If none of these fit because you hire in person, you have no meeting-detection problem to buy for. The hiring integrity stack shows where a detection layer sits against application scoring, the identity gate, and post-offer screening.