Cover deepfake fraud without building the capability.
Extend the engagements you already run, or take the data for work already in flight.
Sell deepfake fraud into accounts you already have.
Penetration testing, red teaming, and social engineering, now covering deepfake fraud.
A fixed-scope deliverable inside your engagement. We do not sell around you.
Our section, co-branded, delivered under your banner.
Already working on it? Take the data.
Labeled real and synthetic attacks, broken out by generator and condition.
Frontier attacks your model has not seen before.
The same corpus under platform conditions, per demographic cell.
Co-authorship where the work warrants it. Otherwise just the data and a citation.
Request research accessDetection vendors: join the benchmark.
An endpoint. Nothing else changes on your side.
Your detector is measured again as new attacks and conditions land.
Public reports carry your name, or anonymize you on request.
Your results across generators, platform conditions, and demographic groups.
Or pull the attack data yourself.
Each card links to the partner's own site.
The questions partners ask.
No. You keep the customer relationship. Our deliverable is a fixed-scope technical evaluation that ships as a co-branded section of your report.
Revenue share or a fixed pass-through, your choice.
The client and the social-engineering scope. We plan the detection-layer evaluation with you under a single statement of work.
Nothing for non-commercial research. The corpus is licensed CC BY-NC 4.0, for evaluation and methodology work rather than training a commercial detector.
Join the benchmark with an endpoint, or pull attack data through the API and run it yourself. A private evaluation of your own model is a service, not a partnership, and it lives on the services page.
Become a partner.
Tell us which of the three you are, and we will scope it from there.