For most of the last two years, whether to tell a customer they were talking to AI was a judgment call. That is changing. A wave of new rules, most notably in the EU and across a growing list of US states, is turning disclosure from a trust decision into a compliance requirement, and the timeline for some of them has already arrived.
This is not a legal advice piece, and the details vary by jurisdiction and by use case. What follows is a plain description of the direction regulation is moving and why support operations should be paying attention now rather than after a rule takes effect.
The EU set a hard date
The EU AI Act's transparency obligations, laid out in Article 50, apply from August 2, 2026. The core requirement is straightforward to describe even if the compliance work behind it is not: AI systems that interact directly with people must be designed so those people know they are dealing with AI. This is not limited to systems classified as high-risk. A business running an ordinary customer support chatbot for EU customers falls within scope simply by virtue of that chatbot talking to people directly.
US states are moving too, unevenly
The United States does not have a single federal disclosure rule, so the picture is a patchwork rather than one standard. California's SB 243 took effect January 1, 2026, but it targets "companion chatbots," AI built to simulate relationships or emotional connection, and it explicitly excludes customer service bots from that specific requirement. Maine's Chatbot Disclosure Act addresses the same kind of question, requiring businesses to notify consumers when they are not interacting with a live human. Colorado has also passed an AI Act, but it is a different kind of law: it addresses AI used in consequential decisions such as lending, employment, housing, and healthcare, rather than chatbot conversation disclosure specifically. More state legislatures have active bills addressing AI disclosure in various forms. The result is that what counts as required disclosure, and which law applies, depends heavily on where a business operates, who it is talking to, and what the AI is doing, and that is unlikely to simplify soon.
Why this is a support design question, not just a legal one
Whatever a given jurisdiction requires, disclosure has to be built into the actual interaction, not bolted on as a policy document nobody reads. That means the support operation itself, the chat flow, the escalation path, the wording a bot uses to identify itself, is where compliance either happens or does not. A rule that says people must know they are talking to AI is, in practice, a requirement about how a conversation is designed from the first message.
What this means for support operations right now
Regulatory timelines are firm even where the underlying rules are still being clarified through guidance. Businesses serving EU customers are working against an August 2026 date already on the calendar. Businesses operating across US states are dealing with requirements that differ by state and by the type of AI involved. Confirming what applies to a specific business, given where its customers are and what its AI systems do, is a question for legal counsel, not a generic checklist. What a support operation can control in the meantime is making sure disclosure is built into the design of every AI-handled interaction, so that whatever the applicable rule turns out to require, the conversation is already built to meet it.
Frequently asked questions
Do I have to tell customers they are talking to a bot?
Requirements vary by jurisdiction. The EU AI Act requires it broadly for AI systems that interact directly with people, effective August 2, 2026. In the US, requirements differ by state and by the type of chatbot involved. Confirm your specific obligations with legal counsel.
Does California's SB 243 require customer service bots to disclose AI use?
SB 243 targets companion chatbots, AI designed to simulate relationships or emotional connection, and explicitly excludes customer service bots from that particular requirement. Other state and federal rules may still apply to customer service use cases.
When does the EU AI Act disclosure requirement take effect?
The transparency obligations under Article 50 of the EU AI Act apply from August 2, 2026, covering AI systems that interact directly with people, including customer support chatbots serving EU customers.