According to Fast Company, a coordinated campaign using AI-generated videos is targeting New York City mayor-elect Zohran Mamdani with fake constituent criticism. The “Citizens Against Mamdani” accounts on TikTok, Twitter, and Instagram feature hyper-realistic videos of imaginary New Yorkers attacking Mamdani’s alleged anti-Americanism and policy plans. These accounts have accumulated tens of thousands of views, with the main TikTok account reaching about 30,000 likes. Cybersecurity expert Emmanuelle Saliba of GetReal Security explains that generative AI has essentially automated astroturfing, making it possible to create convincing fake supporters or detractors without human influencers. The campaign demonstrates how political operatives no longer need real people to spread messages, creating what appears to be organic opposition where none exists.
From paid influencers to fake people
Here’s the thing that really worries me about this. We’ve moved beyond simple deepfakes of politicians saying things they never said. Now we’re seeing the creation of entirely fictional constituents. These aren’t just fake videos – they’re fake people with fake opinions designed to create the illusion of grassroots opposition. And the scary part? It’s working well enough to get tens of thousands of views without anyone immediately realizing they’re completely synthetic.
The undetectability problem
When Saliba says this is “pretty much undetectable without technology,” that should send chills down everyone’s spine. Think about it – regular people scrolling through TikTok or Instagram aren’t running deepfake detection software. They’re just consuming content. The barrier to creating convincing fake opposition has dropped to nearly zero, while the ability to detect it remains firmly in the hands of experts with specialized tools. How many similar campaigns are already running that we simply haven’t noticed yet?
This changes everything for local politics
Local elections have always been somewhat insulated from the worst national political manipulation tactics because they’re smaller, cheaper operations. But now? Any campaign with a few hundred dollars and basic AI tools can generate what looks like widespread local opposition. They can create videos that match local accents, feature diverse “residents,” and hit on hyper-local issues. Basically, we’re looking at the weaponization of AI against democracy at the most fundamental level – the local communities where people actually live.
Where does this end?
I keep wondering – what’s the endpoint here? If campaigns can generate unlimited fake supporters or opponents, does public opinion even matter anymore? We’re heading toward a world where political discourse could be dominated by synthetic voices. And the really terrifying part is that this is probably just the crude early version. Wait until this technology improves over the next election cycle or two. The Citizens Against Mamdani campaign might look like child’s play compared to what’s coming.
