Political Ads for Bots? What That Really Means for Democracy

Political media buyers are talking about spending to “reach AI agents.” That phrase alone should give anyone who cares about democracy a moment of pause.

After all, bots don’t vote. People do.

By 2032, political media buyers predict that as much as 50 percent of campaign ad budgets could be aimed at influencing AI agents and automated systems, out of what is expected to be a $20 billion political advertising cycle, according to findings shared at MediaPost’s Marketing Politics conference in Washington and reported by the outlet.

Why shift spending away from television — historically the favored channel for campaigns — and toward machines? Because machines are rapidly becoming the gatekeepers of what people see, read, and consider important. From social media recommendation engines to search algorithms to AI assistants that summarize news and answer questions, automated systems now sit squarely between information and the public.

Automated Influencers

This is a level-up moment. Campaigns aren’t just trying to persuade voters anymore. They’re trying to influence the systems that decide what information reaches voters in the first place.

That’s a very different — and far less transparent — form of influence.

When advertisers talk about “AI agents,” they aren’t referring to a single type of bot. They mean an entire ecosystem of automated systems: platform recommendation engines, content scrapers, AI assistants, and decision tools that filter and prioritize information.

Some of these systems are designed to behave like diligent research assistants, weighing multiple sources. Others? Not so diligent. Not into research. Happy to shovel the slop of the Internet into your feed. (Remember when Amazon debuted early AI shopping recommendations and you’d see things like, “Because of your interest in gardening, we recommend this book on ancient Egypt”?) Optimization is not the same thing as judgment.

These systems may very well help shape election outcomes by influencing what you see online — and what you never see at all.

Political advertisers are betting that if they can influence what these systems prioritize, they can shape what voters notice, what feels credible, and what quietly disappears from view. The target isn’t the bot’s “opinion.” It’s the information pipeline feeding the human.

At this point in the conversation, someone usually says, “Don’t worry. The AI will be ethical.”

Let’s take a deep breath here before we pounce.

Human ethics and AI “ethics” are not the same thing. Humans weigh consequences, wrestle with moral tradeoffs, and are accountable to communities and laws. AI systems optimize toward goals set by designers and shaped by economic incentives. That is not moral judgment. AI does not understand consequences. At best, it operates with guardrails — which means it can give you directions but fail to see you’re about to drive off a cliff.

So, when campaigns target AI systems, they are not appealing to civic values. They are influencing ranking systems, weighting mechanisms, and distribution pipelines.

This shift also has serious consequences for journalism. If automated systems prioritize scalable, frequently repeated content — as many recommendation and aggregation systems do — then original reporting, especially local and investigative journalism, will struggle to reach the public, unable to compete with sponsored narratives and automated content farms.

What does this mean in the very near future? Less accountability. Less transparency. Even more polarization.

Ask the Candidates

It’s time for both the public and the press to start asking political candidates an important question that is far too seldom asked, if at all: What is your platform on news, media, and technology?

Media and technology are not side issues, even though they are routinely treated as such. They are, quite simply, the operating system of democracy. They influence how voters understand every other policy debate — from healthcare to public safety to the integrity of elections themselves.

Little light is shed on these issues for a variety of reasons. Technology policy is often treated as too technical for voters. Media policy is treated as too boring for campaigns. Platforms have grown too fast for regulation to keep up. There is also an anti-media sentiment that needs to be acknowledged, but we did not magically arrive at this moment with news and media behaving as they do. We are here because of generations of political decisions that shaped today’s media landscape — through deregulation, regulatory inaction, or both — often without public scrutiny.

Before we normalize political influence of AI systems, we should be asking whether our democracy is prepared for persuasion that is automated, invisible, and systemic.