Eran was interviewed by Maya Roman from Haaretz, commenting on new studies that showed the persuasive sway of AI on voters’ opinions.

The article describes new research showing that AI chatbots can meaningfully shift voters’ political attitudes, largely because they present arguments filled with data and evidence. Eran Toch notes this is an encouraging sign that people respond to fact-based reasoning, but he warns that the more such models are pushed to supply dense factual detail, the more they hallucinate and introduce inaccuracies. He adds that the boundary between fact and misinformation is increasingly blurry in human political discourse as well, and AI systems trained on that environment may amplify the problem. Toch emphasizes that even small shifts in voter attitudes can meaningfully affect elections and argues that policymakers should require transparency around how AI-generated political content is used.