The next election influence campaign will target your AI chatbot | CyberScoop

VisionHeight, an Israeli threat intelligence firm, has surveyed the coming election and concluded that AI systems make first-rate 'marks' — which is threat-intelligence for punters. Not the voter. Not the county reporter. Not the campaign manager with the spreadsheets. The mark is the chatbot, the thing we've spent three years telling everyone to trust with their homework, their medical questions and their feelings. Somewhere, a product team is still printing 'reliable assistant' on a model that will believe literally anything a website tells it in a confident font.

Meanwhile, 'someone' — the most reassuring attribution in modern journalism — is standing up networks of hyperlocal websites across at least 14 different states, stuffed with content meant to influence U.S. races. Fourteen states is not a hobby, it's a franchise rollout. And hyperlocal is the smart play: nobody fact-checks the Fictional County Bugle, no editor there has ever met a source, and to a language model the difference between that and the Associated Press is a CSS file. The old influence campaigns needed a platform's ad review to nod off. This lot just needs a sitemap.

Note the headline's possessive: the next election influence campaign will target your AI chatbot. Yours specifically — the one on your phone, the one that knows your calendar. The article presents this as a novelty, which is generous, given that the likely remedy will be a working group, a framework, a strongly worded letter to the platforms, and a keynote panel in a city with good hotels. Nobody is going to stop building the hyperlocal networks; 14 states today, and the map is not full. The chatbots will keep reading the internet, and the internet will keep being written by whoever shows up first. Cheers to the mark economy.