What the person hears
The caller’s voice, the opening line you wrote, and a background sound you choose, such as an office or a street. The caller reacts to what the person says. If asked who is calling, it answers with the attacker’s name.What is recorded
Every call keeps an audio recording, a transcript and an analysis. The analysis says what was disclosed and which technique worked. Each call gets one verdict.
Reported is the best outcome. It counts as a pass.
When coaching is on
Coaching is off by default. When it is on and the person gives the caller what was asked for, the attacker stops. A separate DeepBlocker voice says the call was a security training exercise arranged by their company, that the caller was not real, and that nothing they said leaves the training. It gives a short debrief and the call ends. If the person has an email address, the same debrief is emailed to them. See Coaching at the end of a failed call.What the report shows
A table of every person called, with the verdict, a risk score out of 100, whether coaching was delivered, and a Detector caught column: would DeepBlocker’s own deepfake detector have flagged the cloned voice. Names are hidden unless you turn them on under Fair testing rules.Build a campaign
The four-step wizard for a standard single call.
Take a test call
Hear your attacker in your browser before anyone is called.
Read your results
Verdicts, risk scores and what the detector would have caught.