The 2024 US election is shaping up to be the first major democratic test against AI-enabled disinformation at scale. This week, OpenAI shut down a network of Iranian activist accounts generating fake campaign content against both Trump and Harris—a reminder that foreign interference isn't always partisan. The goal is simpler and more dangerous than vote-flipping: create enough noise, contradiction, and distrust that voters stop believing the process itself is legitimate.
The playbook is now familiar. Cyber actors use social engineering—phishing emails, spoofed links—to breach campaign infrastructure and leak internal communications, often manipulated before release. They generate synthetic video and audio content (deepfakes, robocalls) impersonating candidates. They infiltrate news organizations, as happened when Iranian actors breached media infrastructure in 2020 before the FBI intervened. In France's 2022 election, leaked and altered emails from Macron's campaign dropped during the mandatory silence period, when he couldn't publicly defend himself. The asymmetry was the point.
What makes this cycle different is the speed and scale that AI enables. Mass content generation—false articles, convincing fake videos—used to require resources and time. Now a single actor can flood platforms with bespoke disinformation. The Starbucks CEO may have 170,000 drink configurations; electoral disinformation now has infinite variants, each optimized for a micro-target demographic, each plausible enough to pause someone mid-scroll.
The US legal framework hasn't kept pace. Political campaigns can lie with impunity under First Amendment protection, unlike commercial advertising. Broadcast TV stations cannot reject political ads, regardless of falsehood. Cable channels can—CNN famously rejected Trump spots in 2020—but enforcement is inconsistent and reactive. Proposed bans on deepfakes targeting candidates face a familiar tension: protect democracy or protect speech? Both seem to require sacrificing the other.
The real vulnerability isn't the voting machines. It's the information ecosystem itself. If people believe the election is fraudulent before polls close, turnout plummets, violence becomes rational, and the US's core export—democratic legitimacy—loses value worldwide. That's what Moscow, Tehran, and Beijing are actually competing for. They're not trying to win; they're trying to make winning meaningless.