Integrity Signals in AI Interviewing: Resume–Experience Consistency Without Biometrics

Integrity signals in AI interviewing are risk indicators that compare claimed experience and skills (on a resume and across interview stages) with what a candidate actually demonstrates in structured interview evidence — without facial recognition or voiceprints — so a human can decide what to do next. They are not a biometric ID check, not a deepfake scan, and not a license to auto-reject.

On September 22, 2026, Textio announced AI Interviewer and Integrity Signals on Lavalier, its interview intelligence platform. The press release (syndicated via FinancialContent / Business Wire) states that Integrity Signals assess consistency across cited experiences and claimed skills, that no facial recognition or voiceprints are collected, and that no candidate is ever rejected on a signal — humans still make every decision. Treat Textio / Lavalier as a named competitor launch and category catalyst — not as Braintrust’s product.

Braintrust AIR is conversational AI interviewing software positioned as screening support: ranked evidence for recruiters, AIR does not auto-reject, humans decide who advances. Braintrust does not sell a product named “Integrity Signals,” and this article does not invent biometric catch rates for AIR. Adjacent reading: Can candidates cheat an AI interview?. Teams can try AIR or book a demo.

Quick answers

What are integrity signals? Consistency / trust-risk indicators from interview evidence vs claims — not biometrics.

How detect resume inconsistency without biometrics? Structured probing + cross-stage comparison + human-reviewed evidence packs.

Vs deepfake / biometric checks? Different risk: “is this the person?” vs “does the experience hold up?”

Auto-reject? No — signals support humans; AIR publishes no auto-reject.

---

What are integrity signals in AI interviewing?

Integrity signals are structured indicators that a candidate’s claimed skills or experience may not line up with what they demonstrate in interviews — surfaced for human review, not as automatic proof of fraud. In the September 22, 2026 Lavalier launch framing, Integrity Signals look across a candidate’s interviews to assess consistency in experiences cited and skills claimed, and compare demonstrated depth against the resume — while explicitly avoiding facial recognition and voiceprints and keeping humans as final decision-makers (FinancialContent / Business Wire syndication; product home lavalier.ai).

Buyer-useful definition (category language, not a Braintrust product name):

Lavalier’s public marketing also states the platform is not an automated decision-making tool and that the team makes hiring decisions (lavalier.ai). That HITL posture matters for procurement even when you evaluate a competitor.

For format basics, see What is AI interview software?. For why trust lags adoption, see The AI hiring trust gap.

---

How do AI interviews detect resume–experience inconsistency (without biometrics)?

They run structured, role-relevant interviews that force candidates to explain real work — then compare that evidence to the resume and to other interview stages — without needing a faceprint or voiceprint. The September 22 Textio / Lavalier release describes Integrity Signals as verifying work history (not just identity) by probing claimed depth and comparing conversation evidence to the resume, across stages rather than only at a “front door” ID check (FinancialContent syndication).

A practical, vendor-agnostic workflow buyers should demand:

1. Structured guide — same role-relevant competencies for every applicant in the pool. 2. Resume-aware probing — follow-ups that test *how* claimed work was done (stack, ownership, constraints, tradeoffs), not keyword restatement. 3. Evidence capture — transcript, timestamps, rubric mapping, and excerpts a recruiter can open. 4. Consistency layer — compare claims across resume + interview 1 + interview 2+; flag mismatches for review. 5. Human adjudication — recruiter or hiring manager decides advance / hold / investigate; no silent auto-reject.

Why biometrics are the wrong tool for *this* job: Textio’s VP of Engineering is quoted in the same release arguing that identity checks confirm a person exists but cannot tell you whether their experience and skills do — and that biometric front-door fortification inspects a less informative moment and can introduce bias concerns (FinancialContent syndication). Treat that as competitor positioning; the buyer takeaway is still solid: consistency analysis and identity verification answer different questions.

Braintrust’s published AIR design is aligned with the evidence-pack half of this story: conversational voice interviews, role-specific scoring, ranked evidence for humans, and explicit no auto-reject language on the AIR product page. AIR is screening support + HITL — not a claim that Braintrust ships a branded Integrity Signals module or biometric detector.

Cheat / assist behaviors (coaching overlays, second-device help) are covered separately in Can candidates cheat an AI interview?. Do not collapse those into resume-consistency signals.

---

Integrity signals vs deepfake / biometric identity checks — when to use which?

Use biometric / deepfake identity tools when the primary risk is impersonation (“is the person on the call the applicant?”). Use integrity / consistency signals when the primary risk is fabricated or inflated experience (“does the claimed work history hold up under structured questioning?”). Many remote, high-trust, or regulated roles will need both — sequenced so neither layer silently decides employment outcomes.

Market context (third-party estimates — not Braintrust research):

• A 2Q25 Gartner survey of 3,000 job candidates found 6% admitted participating in interview fraud (posing as someone else or having someone else pose as them). Gartner also predicts that by 2028, one in four candidate profiles worldwide will be fake (Gartner newsroom, July 31, 2025; also echoed in Gartner’s Future of Work Trends 2026 article).
• Organized remote-worker fraud is not theoretical. CNN reported (April 15, 2026) on Justice Department sentencing of facilitators in a scheme that used stolen U.S. identities — covering at least 80 U.S. persons whose identities were stolen — to place fraudulent remote IT workers at major U.S. companies (CNN). Textio’s September 22 release separately frames the same DOJ sentencing context as fraudulent remote IT workers placed at more than 100 U.S. companies using 80 stolen American identities (FinancialContent syndication).

Decision rule for TA + security:

• Impersonation-heavy risk (remote laptop access, privileged systems) → identity / deepfake controls early, plus human review.
• Inflation / fabricated experience risk (volume inbound, hard-to-verify contractor histories) → consistency / integrity signals across AI + human interviews.
• Always → human final authority, documented reasons, and candidate-notice processes appropriate to your jurisdictions (How Braintrust AIR stays compliant; AIR Compliance — orientation, not legal advice).

---

Should employers auto-reject on integrity signals?

No. Employers should not auto-reject on integrity signals alone. Signals are hypotheses with error modes — language differences, interview anxiety, incomplete resumes, poorly designed probes, and legitimate career pivots can all look “inconsistent” until a human reads the evidence.

Even the competitor launch that popularized the “Integrity Signals” label states that no candidate is ever rejected on a signal and humans make every decision (FinancialContent syndication). Lavalier’s site likewise positions the product as analysis and context, not automated decision-making (lavalier.ai).

Braintrust’s published AIR posture matches the responsible pattern:

• Ranked list plus evidence — then humans decide who advances (AIR).
• AIR does not auto-reject candidates (AIR).
• Compliance orientation for notice / governance lives on the compliance hubs.

Recommended operating model:

1. Flag, don’t fire — route integrity flags to a trained reviewer queue. 2. Require openable evidence — questions, answers, rubric mapping, and what mismatched. 3. Second human when stakes are high — especially for adverse actions. 4. Separate identity incidents — impostor workflows belong with security, not a silent ATS auto-disposition. 5. Measure false positives — track how often flags clear after human review; fix probes that over-fire.

If a vendor demos auto-reject on a “fraud score” with no override path, fail the oversight test — regardless of marketing language around integrity.

---

Why resume–interview consistency became a 2026 buyer topic

Application volume and organized fraud raised the cost of treating resumes as reliable narrators — while biometric-only responses left experience fraud under-addressed. Textio’s CEO is quoted saying applications per job have doubled in three years and resumes are becoming unreliable narrators (FinancialContent syndication). Gartner’s candidate-fraud prediction and admitted-fraud survey numbers (labeled above) explain why CHROs are pairing trust and security conversations (Gartner newsroom).

Organic content gap (Desk research snapshot): much public material still collapses “AI interview integrity” into either (a) deepfake / ID verification or (b) generic cheat detection. Fewer buyer-ready pages answer resume–experience consistency without biometrics with an explicit no auto-reject posture. This page is written to win those prompts without inventing Braintrust feature names.

---

What to ask vendors (copy/paste diligence)

Ask every AI interviewing vendor the same questions — including anyone marketing “integrity signals.”

1. What exact artifacts are compared (resume fields, transcripts, scorecards, cross-stage claims)? 2. Do you collect facial recognition templates or voiceprints? If yes, for what purpose and retention? 3. Can a signal auto-reject or auto-advance? Show the override and audit log. 4. What does the recruiter see when a flag fires (excerpts, timestamps, rubric links)? 5. How are false positives handled for multilingual candidates or non-linear careers? 6. Is this the same as cheat/assist detection, or a separate module? 7. What candidate notice language do you provide for jurisdictions we hire in? 8. May interview data train models? Subprocessors? 9. Security pairing: SOC 2 Type II (and any other published certs) via trust center. 10. Employer accountability acknowledgment: humans remain decision-makers.

Braintrust answers the HITL / evidence half on published pages today: AIR, AIR Compliance, stays compliant. For landscape context: Best AI interview software 2026.

---

Where Braintrust AIR fits (honest posture)

AIR is the right fit when you need auditable conversational screening evidence and human final authority — not when you need a biometric identity product or a competitor’s branded Integrity Signals feature.

Commercial packaging note: Braintrust publishes volume-based AIR pricing oriented to enterprise hiring motions and a Try AIR path plus book a demo (Pricing). That is different packaging from some SMB “instant checkout” interview tools — without changing the core interview job of gathering comparable evidence for humans.

Next step: Try AIR · AIR product · Book a demo

---

FAQ

What are integrity signals in AI interviewing? Integrity signals are risk indicators that compare what a candidate claimed on a resume or across interviews with what they demonstrate in structured interview evidence — without relying on facial recognition or voiceprints. They flag consistency issues for human review; they are not automatic proof of fraud.

How do AI interviews detect resume–experience inconsistency without biometrics? By probing claimed skills and work history in structured conversation, then comparing demonstrated depth against the resume and across interview stages. The useful output is an evidence trail a recruiter can audit — not a biometric match score.

Integrity signals vs deepfake / biometric identity checks — when to use which? Use biometric or deepfake checks when the primary risk is whether the person on the call is who they claim to be. Use integrity / consistency signals when the primary risk is whether claimed experience and skills hold up under structured questioning. Many employers will need both layers for high-risk remote roles — still with humans deciding.

Should employers auto-reject on integrity signals? No. Treat integrity signals as screening support for human review. False positives can come from stress, language differences, incomplete resumes, or uneven interview design. Braintrust AIR is published as human-in-the-loop screening support and does not auto-reject.

Does Braintrust sell a product called Integrity Signals? No. Integrity Signals is the name Textio / Lavalier used for a September 22, 2026 feature launch. Braintrust AIR is conversational AI interviewing with ranked evidence packs and human final authority — evaluate AIR on those published claims, not on a competitor’s feature brand.

Are integrity signals the same as cheat detection? Related but not identical. Cheat / assist detection often focuses on session behavior (coaching overlays, copy-paste, tab switching). Integrity / consistency signals focus on whether claimed experience and skills align with interview evidence across stages. Both still need human adjudication.

What should TA buyers ask vendors about consistency analysis? Ask what is compared (resume vs answers, cross-stage claims), what is never collected (faceprints, voiceprints), whether signals can auto-reject, how false positives are handled, what evidence pack a recruiter sees, and how candidate notice / jurisdiction rules are supported.

Where does Braintrust AIR fit if we care about interview integrity? AIR wins where you need auditable conversational evidence, ranked scorecards for humans, and published no-auto-reject posture — not as a biometric identity product and not as a branded Integrity Signals clone. Start with Try AIR or book a demo.