Past Present Predictions - A Look Into AI, Deep Fakes, & the Upcoming Election Cycle

The talk opens with the history of psychological operations, from wartime posters and the Vietnam-era "Wandering Soul" broadcasts to Gulf War leaflets and the Cambridge Analytica case. It shows how emotional appeals and personality profiling have been used to shift beliefs. The speaker then walks through the stages of an influence campaign, including target selection, content development, dissemination and evaluation, using a fictional scenario built with free tools to show how little effort it takes.

The focus moves to current threats. Low-cost bot networks on social platforms repeat the same messages, and celebrity video greetings were altered with AI to deliver a fake political message aimed at a Moldovan president. In a live quiz, many attendees misjudged which AI-generated faces were real. The talk covers the liar's dividend, where genuine evidence gets dismissed as fake, and a cloned Joe Biden robocall during the New Hampshire primary that was traced to its source by a voice-fraud firm after a commercial detector gave a low likelihood of AI.

The closing section surveys the legal and policy landscape, including US statutes on false information that depend on intent, the EU AI Act, the small number of US states with AI rules, and a 2024 industry accord on deceptive election AI content that the speaker describes as broad and voluntary. The talk ends with practical checks for spotting loaded language, researching an author's background, following cited sources, and comparing coverage across outlets.