OpenAI Bans AI Ads for Political Campaigns, But Nobody Is Following the Rules

OpenAI banned the use of ChatGPT for political ad campaigns, but US election teams are largely ignoring the rules. What does this mean for the future of AI-generated election communication?

ChatGPT and political advertising have become a new flashpoint in the AI world this week. OpenAI has explicitly stated in its usage policies that political campaigns are not allowed to use the chatbot to generate advertisements, campaign imagery, or persuasive political messaging. Yet an investigation by the Washington Post reveals that US election teams are widely ignoring this rule. The incident once again puts a spotlight on how much control tech companies like OpenAI actually have over the misuse of their own technology.

What exactly does OpenAI's ban entail?

OpenAI, the company behind ChatGPT, enforces a policy that prohibits political actors from using the tool to draft targeted election advertising, generate persuasive vote-canvassing text, or produce imagery intended to influence voters. The company says this policy is meant to counter election manipulation, disinformation, and the unchecked use of artificial intelligence in democratic decision-making. Similar rules now apply at virtually every major AI provider, from Google to Meta, though the exact boundaries differ per platform.

Campaigns are ignoring the rules anyway

Reporting by the Washington Post shows that multiple US campaign teams, despite the explicit ban, are using ChatGPT to produce ad copy, slogans, and even imagery. According to those involved, this often happens indirectly: text is drafted through ChatGPT under the guise of "research purposes" or "internal communication," only to end up in advertisements afterward. Others use the tool to generate variations of existing ads, making it hard to prove the content was actually created for political advertising. This approach illustrates a fundamental problem with enforcing AI policy: once a piece of text or an image has been generated, it's nearly impossible to trace whether it was produced via ChatGPT. OpenAI does have detection mechanisms and reporting systems in place, but in practice they turn out not to be watertight.

Why this is a bigger problem than it seems

The risk of AI-generated political advertising extends well beyond simply breaking terms of use. Experts have long warned that generative AI drastically lowers the barrier to producing persuasive, personalized propaganda. Where a campaign team once needed an entire communications agency to craft thousands of targeted messages, it can now be done with a handful of prompts. That doesn't just speed up how quickly disinformation can spread, it also expands the scale at which voter microtargeting takes place. This development fits into a broader trend also visible throughout the history of artificial intelligence: whenever a new technology becomes available to the general public, it doesn't take long before it gets used for purposes it was never intended for. From deepfakes to automated fake accounts, the political arena keeps proving to be a testing ground for the limits of AI regulation.

How is OpenAI responding?

OpenAI emphasizes that the policy is being strictly enforced and that accounts violating the rules can be blocked. At the same time, the company acknowledges that full control is nearly impossible as long as users can route around the tool's restrictions. OpenAI says it is exploring additional safeguards, including better detection of political content and stricter verification for accounts that submit an unusually high volume of campaign-related prompts. Critics, however, argue that such measures come too late, with the election cycle already well underway by the time they're implemented. With the 2026 US midterm elections approaching, pressure is mounting on tech companies to establish airtight agreements on the use of generative AI in political communication.

Conclusion: a test case for AI regulation

The situation surrounding ChatGPT and political advertising shows just how difficult it is to actually enforce policy around AI applications, even for a market leader like OpenAI. As long as no watertight technical controls exist, the responsibility largely falls back on users themselves, and that's precisely where things are going wrong right now. For those who want to learn more about how AI is used outside the political sphere, the page on AI applications offers a solid overview. Anyone wanting to follow developments closely can check out more AI news, and for a deeper dive into the underlying technology, our knowledge base is a good resource. One thing is clear: the debate over AI and elections is far from over this fall.

Washington Post

---

Source: Washington Post