# UK Electoral Commission Deploys AI to Counter Election Deepfakes New AI system aims to flag misleading election content within three hours of posting. - **Source:** [Security Affairs](https://brifing.si/source/e6c4f5ec-3e1b-4152-a932-056243bd56eb) - **Published:** 2026-10-19 - **Topic:** [Business](https://brifing.si/c/business) - **Original article:** - **On Brifing:** https://brifing.si/s/695ed6d1-3d05-4c61-aa3a-b54c3621e889 ## Three-Hour Detection Target The UK Electoral Commission is developing an AI system designed to identify and flag damaging deepfakes within three hours of their initial appearance. CEO Vijay Rangarajan emphasizes that this timeframe is critical because it matches the typical speed at which viral content spreads across social media platforms during active political campaigns. ## May 2026 Pilot Results During the May 2026 local elections, the Commission conducted a full-scale pilot program analyzing three million social media posts. The AI flagged 147 pieces of content for human review. Ultimately, ten items were identified as highly misleading, leading to referrals to police or social media companies for further necessary action. ## Focus on Election Integrity The Commission explicitly avoids arbitrating political disputes or campaign claims. Instead, the AI system focuses exclusively on content that threatens the administration of the election itself, such as false information regarding voting procedures or fabricated candidate withdrawals. This narrow scope ensures the regulator remains within its defined legal and operational boundaries. ## Legal and Regulatory Challenges Rangarajan notes that current election laws, dating back to 1983 and 2000, are ill-equipped for the digital information age. The Commission is currently seeking legally enforceable takedown powers from the government to address modern threats. Updating these frameworks is essential to ensure the regulator can effectively manage information-based risks during future elections. ## Algorithmic Bias Concerns Beyond deepfakes, the Commission has detected signs of algorithmic bias on social networks during election periods. Rangarajan described scenarios where platforms might suppress or promote specific political views without warning, potentially distorting voter perception. Currently, UK law provides no clear mechanism to address these platform-level interventions or algorithmic manipulation tactics. ## Cybersecurity and Infrastructure Threats The Electoral Commission faces significant digital pressure, encountering between 30,000 and 40,000 cyberattacks every month. These incidents include botnet attacks and phishing attempts, sometimes peaking at 10,000 daily. Rangarajan advocates for stronger, independent security capabilities to protect the integrity of the UK's electoral infrastructure against these persistent and evolving external threats. ## Limitations of Current Scope The pilot program faced challenges with content labeled as illustrative AI scenes, even when labels were placed at the end of captions. Such content currently falls outside the Commission's intervention criteria. This highlights the difficulty of applying rigid regulatory definitions to rapidly evolving media formats used by political campaigns and candidates. ## Future Election Preparedness With the next general election required by August 2029, the Commission is preparing for inevitable attempts to use deepfakes for disruption. While the current system provides a foundation, the path forward depends on whether lawmakers update legislation to grant the Commission the necessary authority to combat sophisticated digital interference effectively. ## Sources covering this story - https://securityaffairs.com/201307/uncategorized/deepfakes-and-election-manipulation-uk-watchdog-seeks-faster-detection.html — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._