Many US election deniers have spent the last three years flooding local election officials with paperwork and filing thousands of Freedom of Information Act requests to expose alleged cases of fraud. “I’ve had election officials tell me that in an office where there are one or two workers, they were literally fulfilling public records requests from 9 to 5 every day, and then it was 5 o’clock and they went on to their normal election duties. ” says Tammy Patrick, executive director of the National Association of Election Officials. “And that is unsustainable.”
In Washington state, election officials were receiving so many FOIA requests After the 2020 presidential election based on state voter registration data, the legislature had to change the law, diverting these requests to the Secretary of State’s office to ease the burden on local election workers.
“Our county auditors came and testified about how long it was taking to respond to public records requests,” says Democratic state Sen. Patty Kederer, who co-sponsored the legislation. “Processing those requests can cost a lot of money. And some of these smaller counties don’t have the manpower to handle them. “You could easily overwhelm some of our smaller counties.”
Now, experts and analysts worry that with generative AI, election deniers could churn out FOIA requests at an even greater rate, drowning poll workers legally required to respond to them in paperwork and hampering the election process. In a critical election year, when Poll workers face increasing threats. and systems are more strained than ever, experts who spoke to WIRED shared concerns that governments are not prepared to defend themselves against election deniers, and that generative AI companies lack the barriers necessary to prevent people seeking stop poll workers from abusing their systems.
Chatbots like OpenAI’s ChatGPT and Microsoft’s Copilot can easily generate FOIA requests, even referencing state-level laws. This could make it easier than ever for people to inundate local election officials with requests and make it harder for them to make sure elections run smoothly and smoothly, says Zeve Sanderson, director of the Center for Politics and Social Media at the New York University.
“We know that FOIA requests have been used in bad faith before in several different contexts, not just elections, and that [large language models] “We’re very good at doing things like writing FOIAs,” Sanderson says. “Sometimes the point of the records requests themselves seems to have been that they require work to respond to. “If someone is working to respond to a records request, they are not working to do other things like administer an election.”
WIRED was able to easily generate FOIA requests for several battleground states, specifically requesting information on voter fraud using Meta’s LLAMA 2, OpenAI’s ChatGPT, and Microsoft’s Copilot. In the FOIA created by Copilot, the generated text asks about voter fraud during the 2020 election, even though WIRED only provided a generic message and did not ask anything related to 2020. The text also included email and mailing addresses specific to where FOIA requests could be sent.