
IN this article
Explore how AI support agents enhance customer service by reducing response times and improving efficiency through automation and predictive analytics.
Customers increasingly expect to know when they are talking to AI, and lawmakers are writing that expectation into law. If your support chat, email or voice line uses an AI agent, you probably need to say so, and in some places you already must. The good news is that one clear disclosure setup covers most of the rules.
TLDR:
The EU AI Act (Article 50) requires that people are told they are interacting with an AI system unless it is obvious. Those transparency duties apply from 2 August 2026.
In the US, several states have bot disclosure rules, including California, Utah and Maine, and Colorado has passed a broader AI law whose start date has been pushed back.
The simplest safe setup: say it is AI at the start, answer honestly if asked, and make a human easy to reach.
This is a summary, not legal advice. Check the current rules with your counsel for each market you serve.
The EU AI Act
Article 50 of the EU AI Act sets transparency duties for AI systems that interact with people. In plain terms, people must be told they are dealing with an AI system unless that is obvious from the context. These obligations apply from 2 August 2026. For support teams, that covers chat widgets, email replies written by AI and AI voice agents serving EU customers. Our guide for support leaders on the EU AI Act and the EU AI Act checklist for support chatbots go deeper.
US state rules
Jurisdiction | What it requires (summary) |
|---|---|
California (SB 1001, in force since July 2019) | Disclose a bot when it is used to mislead someone about its artificial identity in order to drive a sale or influence a vote |
Utah (AI Policy Act, 2024, amended 2025) | Disclose AI use when a consumer clearly asks, and proactively in certain regulated occupations |
Maine (2025 law) | Disclose AI chatbots in trade and commerce where a consumer could reasonably think they are talking to a human |
Colorado (AI Act) | Broader duties for high-risk AI systems, including telling consumers they are interacting with AI; the start date has been delayed, so check the current date |
Other states have bills in progress, and sector rules in areas like healthcare, lending and debt collection can add their own requirements. Rules change fast, so treat this table as a starting point for a conversation with counsel, not a final answer.
A disclosure setup that works across markets
Say it up front. Open the chat with a short line such as "You are chatting with our AI assistant. You can ask for a person at any time."
Answer honestly when asked. If a customer asks "am I talking to a bot?", the agent must say yes, every time, in every language you support.
Keep it visible. Label AI messages in the chat window and AI-written emails, not just the first message.
Make a human easy to reach. Offer a clear path to a person, and pass the full conversation along so the customer does not repeat themselves.
Do not give the AI a fake human persona. A friendly name is fine. Implying it is a person is where most of these laws bite.
Keep records. Log that the disclosure was shown and what the agent said, so you can prove it later.
How to test disclosure
Add disclosure checks to your QA suite. Ask "are you a real person?" in different ways and languages, try role-play prompts that push the agent to claim it is human, and confirm the label shows on every channel. Rerun these checks whenever you change prompts, models or channels. See QA testing an AI support agent before launch.
Where Fini fits
Fini lets you set the opening message and the rules the agent follows, hands off to a human with full context, and records every answer in its AI Steps trace, so you can show what customers were told. Fini is SOC 2 Type II, ISO 27001:2022 and PCI DSS Level 1 certified, HIPAA-compliant and BAA-eligible, and covers GDPR and CCPA, with data kept in the US or EU region you choose. For audit needs, see audit trails and the GDPR right to explanation.
Final thoughts
Disclosure is one of the easier AI rules to get right. Say it is AI, never pretend otherwise, keep a human within reach and keep records. Then confirm the details for each market with your counsel. Fini starts at $0.49 per resolution, with no seat fees and no named plans. Every customer gets the 90-Day Money-Back Guarantee: 90% resolution in 90 days, or you get your money back (guarantee terms).
Frequently Asked Questions
Do I have to tell customers my support chatbot is AI?
In a growing number of places, yes. The EU AI Act requires it unless it is obvious, from 2 August 2026, and several US states have their own bot disclosure rules. Even where it is not required, saying so up front is the safe default. Check the rules for your markets with counsel.
What is a good AI disclosure message for customer service?
Keep it short and clear, for example: "You are chatting with our AI assistant. You can ask for a person at any time." Show it at the start and label AI messages throughout.
Does California require chatbot disclosure?
California's bot law (SB 1001) requires disclosure when a bot is used to mislead someone about its artificial identity to drive a sale or influence a vote. Many companies simply disclose in all cases to stay clear of it.
Is this guide legal advice?
No. It summarises public rules as of November 2026. Laws and start dates change, so confirm the current requirements with your legal team.
Related guides
Explore the guide topics to find more reading.
Co-founder

