An AI-generated audio clip falsely depicting UK opposition leader Keir Starmer verbally abusing staff was posted on social media during the Labour Party conference, receiving over 1.4 million views before being debunked as a deepfake.
An AI-generated audio deepfake of UK Labour Party leader Sir Keir Starmer was posted on X (formerly Twitter) on Sunday morning during the Labour Party conference in Liverpool. The 25-second audio clip purported to show Starmer swearing at and verbally abusing his staff members. The clip was posted by a pseudonymous account called 'El Borto' (@leo_hutz) that was created in January 2023 and had previously spread unsubstantiated claims about Starmer. The audio received over 1.4 million views and was widely shared across social media platforms including Facebook and TikTok. Analysis by Reality Defender, a deepfake detection company, found the audio to be 75% likely manipulated, with CEO Ben Colman noting that background noise appeared to have been added to evade detection. The British government's own analysis confirmed the audio was fake. Politicians from both major parties, including Conservative MPs Tom Tugendhat and Simon Clarke, quickly denounced the clip as a deepfake and warned against sharing it. The incident occurred just one week after similar AI-generated audio was used to interfere in Slovakia's elections, where a fake recording of an opposition leader discussing election rigging was released before voting. The UK incident highlights growing concerns about AI-generated disinformation ahead of the 2024 general election.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI systems that inadvertently generate or spread incorrect or deceptive information, which can lead to inaccurate beliefs in users and undermine their autonomy. Humans that make decisions based on false beliefs can experience physical, emotional or material harms
AI system
Due to a decision or action made by an AI system
Intentional
Due to an expected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed