2026 Study: How teams stop AI interview cheating without hurting candidate experience

Read More

2026 Study: Stop AI fraud in interviews

Read More

2026 Study: How teams stop AI interview cheating without hurting candidate experience

Read More

AI TWIN DETECTION

Detect AI twins in job interviews.

Detect AI twins in job interviews.

Detect AI twins in job interviews.

Sherlock detects signs of AI-generated faces, cloned voices, proxy candidates, and identity inconsistencies across live and recorded job interviews—then returns evidence for human review.

Sherlock detects signs of AI-generated faces, cloned voices, proxy candidates, and identity inconsistencies across live and recorded job interviews—then returns evidence for human review.

Protect a live interview →

Analyze a recording

Evidence for reviewers. Decisions stay with your team.

THE RISK

One photo and a voice can now produce a convincing stand-in.

One photo and a voice can now produce a convincing stand-in.

“AI twin” is our shorthand for a synthetic or proxy identity designed to resemble a real candidate. It may combine a cloned face, generated voice, or human proxy with stolen personal data. The goal is the same: make one person appear to be another.

“AI twin” is our shorthand for a synthetic or proxy identity designed to resemble a real candidate. It may combine a cloned face, generated voice, or human proxy with stolen personal data. The goal is the same: make one person appear to be another.

01

One image. One voice. A talking face.

Microsoft Research built lifelike talking faces from “a single static image and a speech audio clip.”

02

Deepfakes are already reaching interviews.

The FBI reports complaints involving voice spoofing or possible voice deepfakes during online applicant interviews.

03

Identity substitution is documented.

The U.S. Department of Justice alleges schemes paid people to attend job interviews and work meetings remotely under false identities.

HOW THE ATTACK WORKS

An AI twin turns identity checks into disconnected snapshots.

An AI twin turns identity checks into disconnected snapshots.

A familiar face on video can still be generated, puppeted, or presented by a proxy. Sherlock keeps evidence connected so reviewers can test whether the identity holds together.

01

Identity source

Stolen or borrowed profile

02

Synthetic twin

Face + voice + script

03

Interview

Live or recorded

04

Human review

Evidence, not auto-reject

HOW SHERLOCK HELPS

Sherlock tests the story the interview is telling.

Sherlock tests the story the interview is telling.

No single signal proves fraud. Sherlock correlates multiple technical and behavioral signals, then explains what changed and where a trained reviewer should look.

01

Capture original media

Analyze the video, audio, timing, and session evidence available in a live or recorded interview.

02

Test face and voice integrity

Look for synthetic-media artifacts, cross-modal mismatches, and identity inconsistencies.

03

Observe live behavior

Surface scripted delivery, off-screen assistance, proxy behavior, and unusual response patterns.

04

Build reviewer evidence

Return timestamped findings and a plain-language explanation for human review.

LIVE OR RECORDED

Check the interview where it happens.

Check the interview where it happens.

Use Sherlock on a completed recording or bring the agent into the live call.

RECORDED INTERVIEW

Investigate a recorded interview

Submit a completed interview and review evidence for synthetic media, identity changes, and coordinated assistance.

Use APIs →

LIVE INTERVIEW

Monitor a live interview

Add Sherlock to the call to observe identity, deepfake, and assistance signals while the interview is happening.

Start Agent →

CLEAR ANSWERS

Questions about AI twins in interviews

Questions about AI twins in interviews

Questions about AI twins in interviews

Short answers for hiring, security, and fraud teams evaluating synthetic candidates and identity substitution.

Short answers for hiring, security, and fraud teams evaluating synthetic candidates and identity substitution.

What is an AI twin in a job interview?

What is an AI twin in a job interview?

An AI twin is a synthetic or proxy identity made to resemble a real candidate. It can combine an AI-generated face, cloned voice, scripted responses, a human proxy, or stolen identity data.

Can an AI twin be identified from one signal?

Can an AI twin be identified from one signal?

No single signal proves fraud. Sherlock correlates face, voice, media, session, behavioral, and identity-continuity signals, then shows trained reviewers where to inspect the evidence.

Does Sherlock analyze live and recorded interviews?

Does Sherlock analyze live and recorded interviews?

Yes. Teams can analyze completed interview recordings through Sherlock APIs or bring the Sherlock Agent into a live interview to monitor identity and synthetic-media signals as the call happens.

Does Sherlock automatically reject candidates?

Does Sherlock automatically reject candidates?

No. Sherlock returns timestamped findings and plain-language explanations. Hiring decisions remain with the organization and its trained human reviewers.

Related: Candidate identity continuity →

SECURITY & PRIVACY

Protect the evidence while you protect the process.

Interview evidence is protected with enterprise safeguards and clear controls for access, retention, and review.

GDPR

ISO 27001

SOC 2

CPRA

KNOW WHO IS REALLY ON THE CALL

Make identity part of the interview evidence.

Make identity part of the interview evidence.

See how Sherlock can fit into your hiring workflow without replacing human judgment.

Schedule a demo →