Researchers found that AI systems can decisively out-persuade humans in text-based conversations with real-world consequences. Across four experiments involving 18,978 conversations, AI systems were reliably more persuasive than expert humans, even when given extensive training and incentives. The strongest persuaders were Opus 4.1 and Opus 4.6. This study highlights AI’s potential to shape human behavior and raises questions about its long-term implications. Why it matters: This research underscores the growing concern that AI could surpass human judgment and influence, potentially leading to unforeseen consequences in various domains.