When AI Gets Democracy Wrong: What Happens When Chatbots Mislead Voters

Two House members, one Democrat, one Republican, just sent a letter to federal agencies with a simple message: artificial intelligence is becoming a threat to how Americans vote, and nobody seems to be in charge of stopping it.

The problem is concrete. Chatbots can give voters false information about where to vote, when to vote, or who's on the ballot. A voter in New Jersey asks a chatbot when early voting starts. The chatbot confidently gives the wrong date. That voter shows up too late. Their vote doesn't get cast.

Multiply that by thousands or millions of voters in a close election, and you've got a real threat to democracy itself.

Reps. Josh Gottheimer (D-N.J.) and Mike Lawler (R-N.Y.) sent the letter Tuesday to the departments of Homeland Security and Justice, asking what safeguards exist. The answer, based on what we know, is: not much.

Why This Matters Right Now

We're heading into 2026 midterm elections. AI tools are getting smarter and more available. Most Americans now use chatbots at least occasionally. If a significant chunk of voters rely on AI for election information, and that AI is unreliable, we've got a real crisis on our hands.

This isn't abstract. A voter who can't vote because an AI sent them to the wrong polling place has been disenfranchised just as surely as someone facing a poll tax or a broken machine. The mechanism is different. The harm is identical.

The deeper issue: we're building election infrastructure for 2026 on the fly, without clear rules, without clear accountability. Federal agencies are being asked to figure out what their job actually is.

Where It Fits in the Larger Picture

This connects to something bigger. The Hill's reporting highlights a gap that CGP's platform exists to close: the gap between what technology can do and what government actually does about it.

We have 500,000+ unfilled cybersecurity jobs and our nation's digital infrastructure is largely undefended. We haven't updated major election systems in decades. And now we're supposed to protect voting from AI we don't even fully understand yet.

This is also about power. Who gets to decide what a chatbot tells you about voting? Right now, it's the companies that built the chatbots. No law says they have to get it right. No one's checking. No one's accountable if they get it wrong.

Democracy only works when every citizen can participate, and that means they need real, accurate information about how to vote. Deliberately making that harder, or allowing it to happen through negligence, is a threat to the foundation.