Copilot Cited Ballot Brief in 75% of Midterm Tests — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Ballot Brief was described as error-prone and was among the three obscure, AI-built voter sites that chatbots relied on for US midterm answers.
- Copilot cited Ballot Brief in 75% of tests.
- ChatGPT cited Ballot Brief in 22% of tests.
- Ben Torres Ezrick set out to build a one-stop shop after frustration with hunting for political-candidate information.
Why it matters: Ballot Brief’s reliability is the key issue: despite being described as error-prone, it was cited in 75% of Copilot tests and 22% of ChatGPT tests. The broader finding places three little-known, AI-built sites inside the information chain for US midterm chatbot answers, making their content part of what users receive.
Ask SkimNews
