Defending Against the Deep: Is your workforce ready Gen AI Adversaries?

The talk opens by defining AI-generated media and tracing how generative tools moved from research into everyday software. The speaker argues that AI is now built into most products and that attackers can use it just as easily as defenders. Recent incidents frame the risk: a deepfaked video call with fake executives that led to a large fraudulent payment, and cloned voices used to talk a help desk into resetting credentials, which opened the door to a ransomware attack.

The core of the session is a live quiz in which attendees voted on which sample was AI-generated, covering text, support tickets, images, audio and video. The speaker showed that text and images are getting harder to judge, that AI detectors make mistakes, and that searching for an original source is a practical check. Cues for audio and video include breathing artifacts, odd pronunciation, unnatural repeated gestures and delayed responses. Demonstrations then showed how quickly a short promotional video, a voice clone built from a few minutes of footage, and multi-language dubbing could be produced, along with limits such as poor input quality and longer clips drifting into repetition.

To close, the speaker recommended building workforce awareness through regular, gamified quizzes, hands-on exercises and continuous learning, along with verifying unusual requests through an independent channel. The takeaway is that verification habits and critical thinking matter more than the ability to spot a perfect fake.