Use case

Video Meeting Security

Every face on the call can be synthetic. The wire still shouldn't move.

The attack

In the Arup case, a finance employee joined a video call where the CFO and every colleague was a deepfake — and wired $25 million.

Video conferencing became the trust mechanism that dissolved correct suspicion.

Hiring interviews, board calls, and payment approvals now all run through a channel where seeing is no longer believing.

Real-time face and voice liveness scoring for meeting participants on Zoom and Teams.

Decisions made in meetings — payments, access, offers — still route through the enforced protocol before execution.

Meeting risk signals and post-meeting verifications in one audit trail.

Protocol tip Treat meetings as unverified channels by default: nothing agreed on a call executes until it's verified off the call.

Can deepfakes really run live on video calls?

Yes — real-time face swaps are commodity tooling, and documented frauds have used fully synthetic meetings.

How does detection work during a meeting?

Analysis of video and audio streams for generation artifacts and liveness signals, surfaced as a per-participant score.

What about platforms you can't integrate with?

The protocol layer is platform-independent: the decision is gated even when the meeting can't be scored.

The Callback — a short briefing on new scam patterns and payment controls. No more than twice a month.

Unsubscribe anytime.