TechShorts
All Posts

CX Tech

AI Disclosure Rules for Customer Service Bots and What Experiments Measured

AI Disclosure Rules for Customer Service Bots and What Experiments Measured

Bhavika J

Editorial Team

Since 2 August 2026, providers of chatbots, voice assistants and AI agents that talk to people in the EU have had to tell them they are dealing with an AI. On 28 September, California's governor signed AB 1609, which bars large businesses from passing off a customer service chatbot as a human. The field experiments we found on what AI disclosure does to outcomes involved outbound calls, not inbound support.

What the rules require

Article 50(1) of the EU AI Act requires providers to design systems that interact directly with people so those people know they are dealing with an AI. The exception is where that is obvious to a reasonably well-informed, observant and circumspect person. The notice must be clear and distinguishable, and given no later than the first interaction. Article 99 sets fines for breaches at up to EUR 15 million or 3% of worldwide annual turnover, whichever is higher.

The Digital Omnibus delayed other parts of the Act but left this duty on its original date. The Commission's 20 July 2026 guidelines are not binding, but law firm McCann FitzGerald reads them as treating the "obvious" exception narrowly.

US rules vary by state. This summary is not legal advice.

  • California's 2019 bot law covers online bots used with intent to mislead people about their artificial identity to drive a sale or influence a vote. A clear, conspicuous disclosure avoids liability.
  • AB 1609 covers businesses with more than $500 million in gross annual revenue, according to law firm Fox Rothschild. They cannot present a customer service chatbot as human and must disclose it where a reasonable person could be misled. During business hours, they must make a good-faith effort to connect a customer who asks for a person within 15 minutes.
  • Utah requires disclosure of generative AI when a consumer clearly and unambiguously asks, and proactively only for licensed occupations in "high-risk" interactions, under amendments in force since May 2025.
  • Under Maine's 2025 law, using an AI chatbot that could lead a reasonable consumer to think they are talking to a human, without clear and conspicuous disclosure, violates the state's Unfair Trade Practices Act.

What the field experiments measured

Luo, Tong, Fang and Qu randomised more than 6,200 customers of a financial services company to receive outbound sales calls from chatbots or human workers (Marketing Science, 2019). Undisclosed chatbots were as effective as proficient workers and four times more effective than inexperienced ones at producing purchases. Disclosing the bot before the conversation cut purchase rates by more than 79.7%. Customers who knew were curt and saw the bot as less knowledgeable and less empathetic. Later disclosure, and customers' prior experience with AI, reduced the loss.

Xu, Dai and Yan randomly assigned 11,000 truck drivers on a freight platform to calls from a voice chatbot dispatcher (Management Science, 2026). Disclosing the bot's identity at the start cut the probability of a response by around 11%. Adding interjections and filler words raised it by more than 5.6%, and those features still helped when the bot had disclosed itself.

In a randomised field experiment at a multinational telecom company, Gnewuch and colleagues told some chat customers that humans worked alongside the bot (Information Systems Research, 2024). Those customers wrote in a more human style, so fewer requests could be handled automatically and more went to staff.

Humanlike design has its own measured cost

Crolic, Thomaz, Hadi and Stephen analysed chatbot data from an international telecom company and ran four experiments (Journal of Marketing, 2022). For customers who arrived angry, a more humanlike chatbot lowered satisfaction, their evaluation of the firm and later purchase intentions. Customers who were not angry showed no such effect. The authors trace it to inflated expectations that the bot then failed to meet.

Limits of the evidence

The large effects come from outbound calls where the firm wanted a sale or a reply. We did not find a published field experiment measuring how AI disclosure affects resolution or CSAT in inbound support. Luo's data predates large language model voice agents, and prior AI experience reduced the loss it measured, so its 79.7% is a poor guide to a 2026 queue.

What to check when buying

  • Whether disclosure can be set per channel and jurisdiction, including the first turn of a voice call.
  • Whether reporting splits outcomes by disclosure wording and timing, so a deployment can test variants that stay within the law.
  • Whether the persona implies a human. A staff-style name with no AI label is the kind of setup Maine's law and AB 1609 target.
  • Whether the platform detects requests for a person and logs the time to reach one, which AB 1609 turns into a compliance figure for large businesses.
  • Whether staffing plans allow for disclosure changing customer behaviour. In Gnewuch's study, telling customers people were involved sent more work to staff.

What commonly goes wrong

  • Disclosure that lives only in the terms of service. That is unlikely to meet the EU requirement for clear notice no later than the first interaction.
  • Treating Luo's 79.7% as a reason not to disclose. In the EU that option is gone, and Xu, Dai and Yan found humanlike speech features still improved results after disclosure.
  • One humanlike persona for every contact, complaints included. The Crolic study points the other way for customers who arrive angry.

Sources

  • European Commission, AI Act Service Desk. "Article 50: Transparency obligations for providers and deployers of certain AI systems." Regulation (EU) 2024/1689. https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50
  • European Commission, AI Act Service Desk. "Article 99: Penalties." Regulation (EU) 2024/1689. https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-99
  • European Commission. "Guidelines on transparency obligations for providers and deployers of AI systems." July 20, 2026. https://digital-strategy.ec.europa.eu/en/library/guidelines-transparency-obligations-providers-and-deployers-ai-systems
  • Jones Walker. "Yes, August 2 Still Matters: The EU Approved a High-Risk AI Delay, but Most Transparency Obligations Remain." 2026. https://www.joneswalker.com/en/insights/blogs/ai-law-blog/yes-august-2-still-matters-the-eu-approved-a-high-risk-ai-delay-but-most-trans.html
  • Cooley. "EU AI Act: Transparency Obligations Take Effect 2 August 2026." August 3, 2026. https://www.cooley.com/news/insight/2026/2026-08-03-eu-ai-act-transparency-obligations-take-effect-2-august-2026
  • McCann FitzGerald. "AI Transparency: European Commission's Guidelines on Article 50, Part 1 (Provider Obligations)." 2026. https://www.mccannfitzgerald.com/knowledge/data-privacy-and-cyber-risk/ai-transparency-european-commissions-guidelines-on-article-50-part-1-provider-obligations
  • California Business and Professions Code section 17941 (SB 1001). https://california.public.law/codes/ca_bus_and_prof_code_section_17941
  • Perkins Coie. "I Am Robot: California's New Law Requires Disclosure of Use of Bots." https://perkinscoie.com/insights/update/i-am-robot-californias-new-law-requires-disclosure-use-bots
  • Office of Governor Gavin Newsom. "Governor Newsom signs commonsense legislation to make your life easier." September 28, 2026. https://www.gov.ca.gov/2026/09/28/governor-newsom-signs-commonsense-legislation-to-make-your-life-easier/
  • Fox Rothschild. "California AB 1609 Requires Large Businesses to Disclose AI Customer Service Bots and Provide Human Support." September 2026. https://dataprivacy.foxrothschild.com/2026/09/articles/general-privacy-data-security-news-developments/5118/
  • Wiley. "California Closes Legislative Session with Significant AI and Privacy Developments." 2026. https://www.wiley.law/alert-California-Closes-Legislative-Session-with-Significant-AI-and-Privacy-Developments
  • Perkins Coie. "New Utah AI Laws Change Disclosure Requirements and Identity Protections, Target Mental Health Chatbots." 2025. https://perkinscoie.com/insights/update/new-utah-ai-laws-change-disclosure-requirements-and-identity-protections-target
  • Future of Privacy Forum. "Overview of Utah's 2025 Enacted AI Legislation." April 2025. https://fpf.org/wp-content/uploads/2025/04/Overview-of-Utahs-2025-Enacted-AI-Legislation.pdf
  • Maine Legislature. "10 M.R.S. section 1500-DD: Required disclosure of use of artificial intelligence chatbot to engage in trade and commerce." P.L. 2025, ch. 294. https://legislature.maine.gov/statutes/10/title10sec1500-DD.html
  • Maine House Democrats. "Committee advances Kuhn bill to protect Maine consumers by promoting transparency in use of AI technology." 2025. https://www.maine.gov/housedems/news/committee-advances-kuhn-bill-protect-maine-consumers-promoting-transparency-use-ai-technology
  • Luo, X., Tong, S., Fang, Z. and Qu, Z. "Frontiers: Machines vs. Humans: The Impact of Artificial Intelligence Chatbot Disclosure on Customer Purchases." Marketing Science 38(6). 2019 (foundational). https://pubsonline.informs.org/doi/10.1287/mksc.2019.1192
  • INFORMS. "Artificial Intelligence Can Improve Sales by Four Times Compared to Some Human Employees." 2019. https://www.informs.org/News-Room/INFORMS-Releases/News-Releases/Artificial-Intelligence-Can-Improve-Sales-by-Four-Times-Compared-to-Some-Human-Employees
  • Xu, Y., Dai, H. and Yan, W. "Identity Disclosure and Anthropomorphism in Voice Chatbot Design: A Field Experiment." Management Science 72(1). 2026 (published online 2024). https://pubsonline.informs.org/doi/10.1287/mnsc.2022.03833
  • Gnewuch, U., Morana, S., Hinz, O., Kellner, R. and Maedche, A. "More Than a Bot? The Impact of Disclosing Human Involvement on Customer Interactions with Hybrid Service Agents." Information Systems Research 35(3). 2024. https://pubsonline.informs.org/doi/10.1287/isre.2022.0152
  • Crolic, C., Thomaz, F., Hadi, R. and Stephen, A. T. "Blame the Bot: Anthropomorphism and Anger in Customer-Chatbot Interactions." Journal of Marketing 86(1). 2022. https://doi.org/10.1177/00222429211045687