Toward Meaningful Transparency for AI Chatbots: Disclosing Persuasive Intent Reduces Persuasion

The growing role of AI-generated content and AI-enabled systems in public communication has led regulators to demand clear disclosure of content provenance and AI involvement. But the effects of such disclosures remain uncertain. We test two disclosure approaches in their impact on an AI chatbot’s persuasive appeal. In a preregistered experiment, 1,500 UK adults held a short conversation with a persuasive chatbot about one of 60 policy issues. The chatbot was identical for everyone. We randomized the disclosure that people received: nothing (control), a prominent disclosure that they were interacting with an AI (T1), or that disclosure plus the chatbot’s persuasive intent and instructions (T2). The chatbot shifted attitudes by 12.6 points on a 100-point scale in the control group. The AI-identity disclosure was practically equivalent to no disclosure, with a 13.1-point shift, whereas the additional intent disclosure cut the persuasive effect roughly in half to 6.3 points. It also made participants view the campaign’s methods as less acceptable and support stronger penalties against it. For direct chatbot interactions, transparency about AI identity alone does not meaningfully impact its influence.

Introduction. On 2 August 2026, Article 50 of the EU AI Act (Regulation (EU) 2024/1689) became applicable. Providers of AI systems intended to interact directly with people, such as chatbots, must ensure that users are informed that they are interacting with an AI system. Similar initiatives in other jurisdictions have made the mandatory disclosure of AI content provenance and system identity an increasingly prominent instrument of technology regulation (Wittenberg, Epstein, Berinsky, & Rand, 2024). However, recent studies suggest that AI-disclosure has little or no average effect on attitude change (Boissin, Costello, Spinoza-Mart ́ın, Rand, & Pennycook, 2025; Gallegos et al., 2026). This raises the question whether mandatory disclosures of system identity or content provenance are the right approach to help people respond to automated influence campaigns. A short conversation with an AI chatbot can change political attitudes (Hackenburg et al., 2025; Lin et al., 2025), sometimes with effects that persist for weeks (Costello, Pennycook, & Rand, 2024; Hackenburg et al., 2025).

Discussion / Conclusion. Why does the source label fail? One possible explanation is that people, even without explicit warnings, were aware that they were talking to an AI chatbot. Asked who they had talked with, 98–99% of participants in every arm said an AI chatbot, including the unlabeled control arm. In fact, 29.5% of control participants falsely remembered having seen an AI label. The chatbot provided rich information in favor of a policy in a polished style that is typical for such conversations, which was probably a clear indicator for participants that they were not chatting with a human. Adding a label telling people that the content is AI-generated discloses little new information for most participants. Article 50 does not prescribe a specific interface. In our experiment, however, the disclosure was deliberately prominent, as participants saw a pre-chat card and a persistent banner in both T1 and T2.