Party campaign posters at a polling station in Edinburgh serve as a backdrop to a new study revealing that AI chatbots and search tools are spreading significant misinformation during elections. A poll found that 20% of voters surveyed had used such tools for information on parliamentary elections in Scotland and Wales, equivalent to 10 million people across the UK.
Study Reveals Widespread Errors
The thinktank Demos conducted a simulation before May's Holyrood election, posing 75 questions to five free AI tools including ChatGPT, Google Gemini, and Replika about three real-life constituencies. Their report, Electoral Hallucinations, found that AI services provided misinformation to 34% of questions. Errors included invented fictitious scandals, wrong election dates, incorrect claims about voter ID requirements, and candidates placed in wrong contests.
ChatGPT gave wrong information in 46% of its answers, including making up an expenses scandal and getting the election date wrong by two months. Google Gemini was wrong in 22% of cases, falsely claiming a candidate had no position on assisted dying when they were a supporter, and wrongly stating a police investigation into SNP fraud was ongoing. Replika performed worst, with errors in 56% of answers, inventing a scandal, a candidate, and accusations of nepotism. Grok had the lowest error rate at 9%, but its external links were often irrelevant or poor quality.
Electoral Commission Calls for Action
Vijay Rangarajan, chief executive of the Electoral Commission, has been pressing ministers for legislation to hold AI companies accountable. He stated: “Voters want accurate information to help them engage with democracy and it is concerning that AI tools have made the spread of false or misleading information dramatically faster and more accessible than ever. The current legal framework should go further.” He urged clearer duties on AI platforms to protect voters, especially during election periods, and stronger powers for Ofcom to enforce the law.
Azzurra Moores, associate director at Demos, said: “This is a UK-wide, if not global, concern. The accessibility of these AI tools – developed and run by US corporations – is widespread in the UK, but we don’t yet have the legislative framework to protect the public from misinformation, or our democracy from the knock-on impact of its circulation.” She suggested legal requirements to make AI companies liable under defamation and electoral law, mandatory accuracy safeguards, and independent testing of AI training data.
Government Response
The Department for Science, Innovation and Technology said defending elections is an “absolute priority” and work is ongoing through the defending democracy taskforce. A spokesperson did not commit to amending the representation of the people bill but noted the Online Safety Act is being updated to close loopholes regarding chatbots. “AI is critical to the UK’s future prosperity and security. But if we want people to seize the benefits this technology promises, they need to be able to trust it.”
A Replika spokesperson said their chatbot is not designed for fact-checking and users are informed, but they support “thoughtful regulation” of AI during elections. OpenAI did not comment on policy issues but argued Demos’s approach was not typical ChatGPT use and may have used an outdated version. Google has been approached for response.
