US 2026 Election Campaigns Violate OpenAI Rules for AI Political Ads
Artificial intelligence (AI) is infiltrating the political machinery of the 2026 United States midterm elections. A number of candidates from major parties have begun to acknowledge in their campaign finance reports that they are using the technology to reach voters, despite strict limits imposed by leading AI companies to protect electoral integrity.
An analysis by The Washington Post of campaign finance disclosures found that 39 congressional candidates reported payments for OpenAI subscriptions during this election cycle. Two of them explicitly stated in official documents that they used the subscription for advertising, even though OpenAI’s policy strictly prohibits candidates from using its tools to create political advertisements.
Congressman Mike Lawler (Republican), who is running for re-election in a suburban New York district, was recorded as the candidate with the largest OpenAI expenditure since 2025, at approximately US$3,260 (around Rp50.5 million) for ChatGPT. Lawler’s spokesman, Ciro Riccardi, defended the move by describing it as an effort to improve team productivity.
In addition to individual candidates, around 30 Political Action Committees (PACs) and parties also reported payments to OpenAI. The Republican National Committee (RNC) was the largest political spender, with a value of approximately US$9,700 (around Rp150 million).
Eric Wilson, a Republican digital strategist, revealed that AI is widely used to write emails, advertising copy and message scripts. However, almost no one is willing to admit it openly. “We know it is happening, but no one is talking about it,” Wilson said.
This lack of transparency reflects a paradox: generative AI is becoming increasingly prevalent, and candidates feel they will be left behind if they do not use it. On the other hand, studies show that voters tend not to trust messages that are known to have been created by AI.
Katie Harbath, CEO of Anchor Change and a former Meta executive, predicted that as election day approaches, more campaigns will shift AI ad creation to outside parties such as super PACs to avoid direct responsibility, similar to the way they handle negative advertising.
OpenAI has stated that its tools should only be used for responsible internal tasks such as research, planning and administration. However, field testing shows inconsistent enforcement of the rules.
In one experiment, ChatGPT was at one point willing to create a micro-targeted fundraising message for a specific demographic (such as mothers), which clearly violated company policy. At a different time, however, the system refused a similar request. This inconsistency allows campaigns to continue seeking loopholes in OpenAI’s election rules.
Researchers are concerned that widespread use of AI will usher in a new era of faster and cheaper disinformation. AI enables highly personal microtargeting, meaning millions of different messages can be sent to individual voters specifically.
Although more than 30 US states have passed laws restricting the use of deepfakes in campaigns, resources for law enforcement remain very limited. The biggest challenge is no longer simply AI’s ability to create fake images, but rather how this technology is changing the way citizens obtain information and become informed citizens.