Electoral Commission launches pilot to detect deepfakes ahead of May elections

The pilot, which is now live, will operate in the lead‑up to the May elections in England, Scotland and Wales, and is focused on identifying artificially generated audio and video material intended to mislead voters or misrepresent candidates. The Commission has indicated that the initiative reflects growing concern about the speed and sophistication with which false or misleading content can be produced and disseminated using artificial intelligence tools.

According to the Commission, the system will monitor online platforms for content such as fabricated videos suggesting a candidate has withdrawn from an election, or manipulated audio purporting to show a candidate making offensive or inappropriate statements. Such material, while false, may have the capacity to influence voter understanding of the electoral process or undermine confidence in candidates and democratic institutions.

The Commission has pointed to evidence from recent elections illustrating the scale of the challenge. During the 2024 UK general election, more than half of voters surveyed reported encountering misleading information about parties or candidates, while approximately a quarter said they had seen or heard content they believed to be a deepfake. International examples have also been cited, including a 2025 Irish election in which a deepfake video falsely depicted a presidential candidate withdrawing from the race shortly before polling day.

Commenting on the launch of the pilot, Vijay Rangarajan, Chief Executive of the Electoral Commission, said that voters are entitled to accurate and reliable information when participating in elections, and that deliberately misleading or abusive content involving candidates should not be circulated. While noting that deepfakes have not yet materially affected the outcome of a UK election, he emphasised the Commission’s determination to ensure this remains the case.

Mr Rangarajan explained that the pilot is intended to enable the Commission to identify deepfake material swiftly, understand its reach and potential impact, and work with political parties, candidates and online platforms to secure the removal of misleading content or ensure that accurate corrective information is made available. The overarching aim, he said, is to provide voters with confidence that information about how elections operate is trustworthy.

The pilot is being delivered in partnership with the Accelerated Capability Environment (ACE), an innovation unit within the Home Office. The Electoral Commission will be responsible for monitoring content and collating evidence throughout the election period.

Where content is identified that raises serious concerns, the Commission has powers to refer material to the police or other relevant authorities, and may request that social media platforms remove content that is demonstrably false or misleading.

The Commission has confirmed that it intends to publish findings from the pilot following the conclusion of the May elections. These findings are expected to inform future approaches to countering AI‑driven misinformation in the electoral context and may contribute to wider discussions about safeguarding democratic processes in the digital age.