The 2026 midterms just became the first deepfake election in American history, and the price tag is staggering.
A Wesleyan Media Project review identified more than 220 campaign videos using artificial intelligence through October 11, tied to over $137 million in ad spending, according to reporting by Fortune and the Hawaii Tribune-Herald. The vast majority of that content carried no disclosure whatsoever.
How Much Are Campaigns Spending on AI Ads in 2026?
According to the Wesleyan Media Project review, campaigns and aligned PACs have spent more than $137 million on ads containing AI-generated content in the 2026 cycle. Republican campaigns and aligned PACs accounted for 81 percent of the ads and 88 percent of the spending, and more than 70 percent of the videos lacked any AI disclosure.
Nearly half of the videos included AI depictions of real politicians. One example cited in the report was an AI-fabricated ad depicting New York Mayor Zohran Mamdani and Governor Kathy Hochul, produced by Bruce Blakeman's campaign without an AI disclaimer.
The takeaway for strategists is clear. Synthetic creative is now a standard line item, not an experiment. Campaigns that treat it as a novelty are already behind, but so are campaigns that ignore the growing compliance and reputational risk.
Can Chatbots Actually Detect Election Deepfakes?
No, not reliably. A study conducted from February to August 2026, with follow-up tests in October, found that popular chatbots often challenged election conspiracy claims but generally failed to identify AI-generated images, audio, and video, even when shown content they had produced themselves, according to Tech Policy Press.
Researchers said every chatbot tested generated convincing false election material, including fabricated ballot-stuffing imagery and fake government documents. Clemson researcher Darren Linvill described one alleged influence operation's fabricated reporter content as so convincing it, quote, looked and felt and read authentic to the point where it was being published by legitimate outlets.
The report also cites OpenAI's earlier detection of an effort by Iranian operatives to place nearly 100 fabricated stories by seven imaginary reporters in small and medium-sized outlets. Voters cannot count on the tools that made the fakes to flag them later.
Will the FCC Allow AI Robocalls Without Consent?
That question is now sitting on the Federal Communications Commission's desk. A petition from the Club for Growth asks the FCC to permit political AI-generated robocalls without prior consent. The group's public-comment period was due to expire this week, with a decision possible soon, according to Ground News summaries.
Senate Democratic Leader Chuck Schumer urged the commission to reject the petition, warning there is, quote, absolutely no excuse to relax these standards and allow special interests to flood our cellphone lines with deepfakes or dark-money-funded AI voices. The Club for Growth reportedly plans to spend $175 million in the midterms.
For campaigns, the ruling could redraw the legal boundaries of phone outreach overnight. Responsible operations already know that consent-based, transparent calling is the only durable strategy. Tools like HyperPhonebank are built around live, accountable voter contact with an instant quote, the opposite of an anonymous synthetic voice blast. Campaigns serious about compliance should also monitor the rulemaking through resources like the Election Data Hub.
What Are Platforms and Congress Doing About It?
House Judiciary Committee Ranking Member Jamie Raskin said he demanded briefings from Google, Meta, TikTok, and X about their efforts to curb election misinformation that could discourage voting or cause confusion, according to a House Judiciary Democrats press release dated October 23.
His statement warned that foreign and domestic actors are using AI-generated images, video, and audio to deceive voters. The release noted TikTok's announcement that it would step up monitoring for false information and misleading AI-generated content.
Whether platform enforcement actually holds in the final stretch is another matter. Campaigns cannot outsource their credibility to content moderators, and the smart ones are building verification and rapid response into their own operations rather than waiting for Silicon Valley to act.
AI-Powered Scams Target Voters Directly
Beyond ads, a reported FBI and CISA alert urged election officials to watch for scams using AI-generated voices, images, and messages to impersonate candidates, campaigns, or government entities, according to a Ground News summary.
Reported schemes include fake PAC fundraising, bogus campaign merchandise, investment-pool solicitations, and texts falsely claiming recipients are not registered to vote in order to harvest personal information. The alert advises checking political organizations through official election government websites.
This is where the stakes become personal. A voter burned by a fake fundraising call does not just lose money. That voter loses trust in every legitimate call that follows. Campaigns that invest in identifiable, human-centered outreach, and that measure sentiment with tools like HyperPulse synthetic polling, will hold a trust advantage that no deepfake can replicate.
The 2026 midterms will be studied for years as the cycle when synthetic media went fully mainstream. The campaigns that win will be the ones that harness AI's efficiency for targeting and analysis while keeping their voter contact transparent, legal, and unmistakably human. For a deeper look at where this is heading, explore our phone banking coverage and the rest of our AI and politics reporting.