A study published July 1, 2026 in PLOS One by Steffen Herbold of the University of Passau, Germany, and colleagues found that AI-generated impersonations of political figures were judged by members of the public to be more authentic, relevant, and coherent than the speakers’ actual debate responses. Researchers used GPT-4 Turbo to generate impersonated responses to audience questions from 30 episodes of BBC1’s Question Time, prompting the AI with Wikipedia biographies of 112 public figures. A representative sample of 948 UK adults rated both original and AI-generated responses, with differences statistically significant in every comparison. Despite measurable linguistic differences—including a greater range of vocabulary and fewer epistemic markers like "I think" in the AI text—these did not affect authenticity judgments. Around half of the responses considered the content to differ, with further analysis suggesting the AI-generated response addressed the question while the real speaker did not, or expressed different stances. The study examined a single debate format from one country using one AI model, and the authors conclude that AI can generate impersonated political content rated as more authentic than the real thing, raising concerns about targeted misinformation campaigns.