THE UK Electoral Commission is piloting an AI-driven system designed to alert candidates to significant deepfakes within about three hours of posting. At a conference panel, chief executive Vijay Rangarajan outlined that the project aims to help voters, journalists and political parties respond quickly to material that could disrupt the election process, rather than adjudicating political claims.
In a local elections pilot run earlier in 2026, the Commission analysed roughly three million social media posts with AI, flagging 147 items for human review. Ten of these were judged notably misleading and referred to social media platforms or the police; officials emphasised that the documented posts had limited electoral impact in that trial.
The system concentrates on real‑time identification and review rather than identifying authors, with AI scanning posts and flagging suspicious material for human assessment. If criteria are met, the Commission would alert candidates or parties within about three hours of first appearance online—roughly the time it takes for content to go viral.
Rangarajan flagged legal and regulatory hurdles, noting UK law drafted decades ago may not clearly apply to modern AI-enabled misinformation and that takedown powers are being considered. He also suggested potential algorithmic bias on some platforms during elections and said stronger, independent security capabilities might be handled by another body.
The Commission reports it faces thousands of cyberattacks monthly and is seeking to scale secure monitoring ahead of the next general election, due by August 2029, or sooner if called earlier.