Poisoning the Well: Election Deepfakes and the Erosion of Shared Truth

Fabricated audio and video of candidates have moved from a theoretical worry to an operational tactic in the 2026 US midterm cycle. But the gravest damage isn't any single convincing fake — it's the slow poisoning of the well, the point at which voters stop trusting any political information at all. And the rules meant to govern this arrived late, incomplete, and gridlocked.

What Happened

AI-generated deepfakes — fabricated audio and video of candidates — have emerged as a widespread tactic in the 2026 US election cycle, turning what began as a theoretical warning into an operational reality. Meanwhile, the federal response has stalled. The Federal Election Commission remains divided along partisan lines and has not established clear guidelines governing AI use in campaign advertising. The Federal Communications Commission has banned AI-generated voices in robocalls, but that rule does not extend to digital ads, television, or social-media content — leaving the largest channels uncovered.

Advocacy groups have pressed regulators to act within existing authority: Public Citizen petitioned the FEC to affirm that long-standing prohibitions on "fraudulent misrepresentation" extend to deepfake campaign ads, and the Campaign Legal Center urged the commission to use the powers it already has. In Congress, a bipartisan proposal — the AI Fraud Deterrence Act — sought to expand penalties for AI-enabled fraud and criminalise impersonating federal officials with AI. But with no coordinated federal framework in place, analysts describe the cycle as a regulatory vacuum, and experts caution that the objective of political deepfakes is not only to deceive in a single instance but to gradually erode public trust in all political information.

Why It Matters

Elections depend on a shared set of basic facts. When any clip can be fabricated and any real clip can be dismissed as fabricated, that shared foundation weakens — and the damage does not require a single deepfake to succeed. A targeted candidate must spend time and resources debunking a lie rather than making their case, and by the time a fabrication is disproven it has already spread, with reputational harm that may be irreversible. The deeper cost is collective: a public that cannot agree on what is real is harder to inform, harder to persuade with evidence, and easier to divide. This is disinformation aimed less at winning an argument than at dissolving the ground arguments stand on.

advertising