01
Identity at intake
KYC-grade verification via Persona / Stripe Identity / Onfido. Government ID + selfie liveness. Phone and email carrier check. Most fraudulent signups bounce here.
The Carevoices Panel
The verified clinician panel built for AI-moderated research
License + NPI verified at intake. The 6-Layer Fraud Detection Stack including the AI-on-AI Detection Layer. Built for AI-moderated voice and video research, not retrofitted from a survey panel.
License + NPI verified at intake
6-Layer Fraud Detection Stack
0.4% verified AI-respondent leak rate (Q3 2025)
Live
Intelligence Report Live
0% Verified
AI Insight
0.4% verified AI-respondent leak rate (Q3 2025). 6-Layer Fraud Detection Stack. Source: Panel Fraud Transparency Report Q3 2025.
Carevoices
Benchmark
99.6%
Live
Trusted by teams at
The 6-Layer Fraud Detection Stack
Six layers of detection, including AI-on-AI
Built for the AI-respondent era. PNAS 2025 research shows AI bots evade legacy fraud detection 99.8% of the time. The 6-Layer Fraud Detection Stack treats this as a built-in design constraint, not a bolt-on feature. Layer 4 — the AI-on-AI Detection Layer — is the differentiator that legacy text-based detection structurally cannot match.
02
License verification
State board database cross-check for nurses (RN/LPN/NP licenses). NPI Registry validation for physicians. Specialty + practice setting + employer attestation captured at intake.
03
Voice baseline + behavioral fingerprint
10-second voice baseline at intake stored as biometric anchor. Behavioral fingerprint built across interviews — vocabulary, response patterns, hesitation cadence, opinion stability. Drift = flag.
04
AI-on-AI fraud detection
Our AI moderator probes mid-interview for AI-assisted respondents. Dynamic challenges ("repeat that with the word 'mango' in it"), off-topic personal context probes, real-time TTS artifact detection.
05
Payment infrastructure as deterrent
Stripe Connect / Plaid bank verification at first payout ties identity to real money. Bot farms cannot easily monetize when payment is verified.
06
Reputation + economics
Long-tenured panelists earn 1.5–2× rate (reputation premium). Fraud detection = permanent ban + forfeit. High-friction intake deters bots; reward retention deters churn.
Panel FAQs
Common questions
Three structural differences: (1) license + NPI verification at intake (not just at recruitment); (2) behavioral fingerprinting across interviews to detect AI-assisted respondents — a capability legacy survey panels structurally cannot match; (3) panel curation specifically for AI-moderated voice and video research, not for survey breadth. We don't compete on raw panel size.
Default-blinded studies — your team never learns participant identity, so transfers of value typically don't trigger reporting. For unblinded engagements, we capture NPI, state license, specialty, and address at intake and provide CMS-format export data for your sponsor's filing. See /compliance/ for full documentation.
PNAS 2025 research shows AI bots evade fraud detection 99.8% of the time at legacy panels. Carevoices is purpose-built for the AI-respondent era — our 6-layer fraud detection includes the AI-on-AI capability that legacy survey panels structurally cannot match. We publish quarterly Panel Fraud Transparency Reports with verified AI-respondent leak rates.
Yes. Licensed nurses, NPs, PAs, physicians, and specialists can join the waitlist at /panel/join/. We onboard new panelists in waves matched to active study demand. Honoraria are calibrated to specialty rarity and time commitment; rates are published on the panelist portal at intake.
Talk to us
Get a demo
See sample panel feasibility for your specific study question. Walk through compliance posture and verification methodology on the call.
For licensed clinicians
Honoraria scaled to your specialty. Same-day pay. License + NPI verified.