Put an independent number on your detector.
Whether you build detection or buy it, the question is the same: what does it actually catch under adversarial conditions? We answer it two ways, on the same dataset and the same methodology.
Two ways to red-team a detector.
Both engagements run on the same dataset and the same methodology. They differ in who initiates them, who owns the customer relationship, and what the report says on the cover.
01
You submit a detector. We red-team it.
An independent red team against your own model. We attack your detector with the fraud that is actually circulating, under the conditions a real platform imposes, then return a verdict: pass, conditional, or fail. Where it breaks, you get the exact recipe that beat it, so you can fix it.
- Initiated by
- The detection vendor
- Duration
- 4 to 6 weeks, fixed scope
- Deliverable
- Red-team report with a verdict, per-group results, and the recipes that broke it
- Used for
- Procurement evidence, marketing-claim validation, pre-release QA
02
Your red team brings us in for the technical layer.
A partnership with red-team and security-awareness firms. The partner runs the engagement and keeps the customer; we add an independent review of the detection technology in scope, so the end customer leaves with one joint report covering both the human and the technical layer.
- Initiated by
- A red team or security-awareness partner
- Duration
- Matches the host engagement
- Deliverable
- Single joint report covering human and technical layers
- Used for
- Enterprise security audits, joint customer engagements
Why not just use an internal team or a generalist pentest?
Both have their place. Neither is an independent adversary with a deepfake-specific corpus and per-group reporting. That gap is what we fill.
| Capability | Margen | Internal red team | Generalist pentest |
|---|---|---|---|
| Independent of the vendor under test | |||
| Deepfake-specific attack corpus | |||
| Per-group fairness breakdown | |||
| Platform-realistic conditions | |||
| Pre-registered, reproducible method | |||
| Hands back the breaking recipe |
From a free benchmark to a standing red team.
Most buyers start with the public benchmark, then commission an evaluation of the detector they actually run. The white-glove tier is for teams that need ongoing, tailored coverage.
Public benchmark
Free
See how the market's detectors hold up.
- Our published results on leading open-source detectors.
- Per-group breakdowns and platform-condition drops.
- A starting point for narrowing a shortlist.
Evaluation
Per engagement
An independent grade for one detector.
- Your model run through our full benchmark and attack suite.
- Pass, conditional, or fail across groups and conditions.
- The recipe that broke it, where it fails, for fixing.
White-glove red team
Custom
Ongoing, tailored adversarial coverage.
- A sequestered attack set built for your exact threat.
- Co-delivered alongside your red-team or security partner.
- Repeat engagements as new generators emerge.
See how we engage with a team like yours.
The engagement is the same shape for everyone, but the threat, the conditions, and the report change by domain. Open the breakdown for your part of the market.
Detection vendors
Builders that need third-party validation.
See the breakdownBuyerIdentity verification and KYC
Liveness-flow operators under continuous attack.
See the breakdownBuyerHiring and interview platforms
High-volume detection, real candidates, real funnels.
See the breakdownPartnerRed team and security-awareness firms
Extend social-engineering into the technical layer.
See the breakdownBuyerEnterprise security leadership
CISOs procuring detection technology directly.
See the breakdownTell us which mode fits.
Submitting a model for evaluation, or extending a red-team engagement to the detection layer? Either way, start here.