Deepfake and voice-fraud intelligence from the people building the detection. Real attacks, new generation models, what defends against them, and where the regulation is heading.
Why regulated firms need to test how people verify urgent requests, not merely whether they recognise suspicious wording.
Read the post →Deepfake calls are making voices, faces and caller IDs unreliable, so regulated firms must test the workflows that sit behind verification.
Read the post →The reported disruption to the Bankruptcy Noticing Center is a reminder that vishing readiness belongs in operational resilience, not only security awareness.
Read the post →The reported Fideuram incident is a reminder that payment controls must withstand convincing identity, voice and urgency together.
Read the post →AI voice cloning makes familiarity a weak control, so regulated firms need evidence that staff can follow independent verification under pressure.
Read the post →A convincing face and familiar voice should never be able to bypass callback, approval, separation-of-duties and escalation controls.
Read the post →A reported voice-cloning fraud case is a reminder that resilience depends on workflow design, escalation and evidence, not on asking staff to judge whether a voice sounds genuine.
Read the post →Office background noise hid a cloned voice from our own detector. The cause was how detectors average a call, not the model. Here is the test, the fix, and what to check in yours.
Read the post →The evidence is moving from theoretical threat to board-level risk, but many firms still test training and identity checks rather than the phone workflows attackers are targeting.
Read the post →Two stories, eleven days apart, that belong together, and the enforcement gap between them.
Read the post →An open-source app clones a voice in 646 languages from three seconds of audio, locally and free. The threat model just changed.
Read the post →