We Need to Ban AI in Campaign Ads
During the recent New York mayoral debate, Andrew Cuomo’s campaign briefly posted an AI-generated attack ad and then deleted it. But the damage was already done. The video showed a synthetic version of his opponent, Zohran Mamdani, running through the city as fictional “criminals” rallied for him: shoplifters, abusers, and traffickers, all cheering for the leading candidate in the NYC mayoral race.
In one image, a Black man in a keffiyeh steals from a store. In another, a burning skyline suggests chaos under Mamdani’s leadership. The ad was offensive, racist, Islamophobic, and entirely fake. Cuomo’s campaign blamed “a junior staffer” for an accidental upload, but there is no accident in training an algorithm to generate racist caricatures. There is no accident in feeding a machine the biases you do not want to say out loud yourself.
This is what AI allows in politics: the outsourcing of prejudice.
AI does not invent hate from nothing. It learns it from us. Every dataset, every prompt, every image it produces comes from a world already built on inequality. When those systems are used in campaign messaging, they do not just reflect bias, they magnify it.
The Cuomo video did not show real people. It did not use real footage. But it mirrored centuries of racist imagery: the Black criminal, the violent foreigner, the Muslim threat. It blended those stereotypes into one algorithmically polished smear and presented it as truth.
That is the danger of AI in campaign ads. It is not just about misinformation. It is about who gets dehumanized most easily when the technology is unleashed. AI does not distribute harm equally. It hits marginalized communities first and hardest. It makes bias scalable.
AI lets campaigns do three things they have always wanted, faster and cheaper and with fewer fingerprints.
Campaigns can generate dozens of inflammatory clips in a single day, each designed to provoke fear or disgust. The most racist and divisive ones are often the ones that go viral.
Microtargeting allows campaigns to quietly distribute different versions of a video to specific audiences. White suburban voters may see one story, while others see another.
Get Estelle McLaughlin’s stories in your inbox
Join Medium for free to get updates from this writer.
When backlash arrives, campaigns can claim the ad was an accident or the work of a rogue intern. Because the footage was not “real,” they can shrug off the consequences as being just how AI decided to interpret the prompts.
That plausible deniability is what makes AI propaganda so dangerous. It allows politicians to launder discrimination through a neutral-sounding tool.
Developers like to say that AI does not see race. That is exactly the problem. It does not see context, history, or power either. When fed biased data such as mugshots, skewed news images, and surveillance footage from over-policed communities, it learns to replicate those patterns with frightening accuracy.
So when a campaign uses AI to make an “edgy” ad, it is really breathing new life into old stereotypes. It teaches voters that Blackness equals danger, that Muslim equals extremist, that immigrants equal chaos. It allows the algorithm to say what the candidate cannot.
Some argue for regulating AI political ads instead of banning them. They suggest labeling each video with disclaimers like this video contains synthetic media. That is not enough. No voter pauses a TikTok or YouTube ad to read a warning label.
We need a full ban on AI-generated imagery, audio, and video in campaign advertising. Not later. Not after the next scandal. Now.
Once these tools are normalized, they will never go away. Campaigns will claim they need them to compete or to stay modern. But democracy does not need modernization. It needs integrity.The Cuomo ad may have been deleted, but its echoes remain. Screenshots circulate online. The racial tropes it used will keep appearing, training the next model, shaping the next narrative.
If we allow AI to rewrite political reality, we will be letting machines hard-code prejudice into our elections. We will be letting campaigns hide their racism behind pixels and prompts.
AI does not just distort truth. It distorts empathy. It replaces lived experience with simulation and moral accountability with excuses.
And once democracy becomes a hallucination, what is left to vote for?
We still have time to act. Let us say no to AI in campaign ads, not because we fear technology, but because we still believe people deserve to see the truth before they vote.
