Do AI Calls Have to Say They Are AI?
When an AI voice agent must say it is AI: the FCC 2024 ruling, Utah SB 226 and EU AI Act Article 50, with a disclosure line you can use on real calls.
Kurze Antwort
It depends on who you call and where. In the US, the FCC treats AI voices as artificial voices, so consent is required; a federal AI-disclosure rule was only proposed. Utah requires disclosure when asked, and at the start in regulated high-risk cases. The EU AI Act requires people to be told they are dealing with AI unless it is obvious. Disclosing at the start is the safest default.
If you run an AI voice agent on outbound or inbound calls, one question comes up before launch: does the agent have to say it is not a person? The honest answer is that there is no single global rule. Some laws require consent for AI voices, some require disclosure on request, and some require the person to be told without asking. This guide goes through the primary texts we checked, what each one says, and a disclosure line that works on a real call.
It is general information, not legal advice. Rules change, so every section names the source and the date it was checked.
Short answer: it depends on who you call and where
| Wo | Regel | What it says about AI voices |
|---|---|---|
| US, federal | TCPA, FCC 24-17 (2024) | AI-generated voices are an artificial voice; prior express consent is needed unless an emergency or exemption applies |
| US, federal | 47 CFR 64.1200(b) | Artificial-voice messages must state who is calling at the start and give a call-back number |
| US, federal | FCC 24-84 (proposal) | Would require AI-call disclosure at the start; proposed in the document checked |
| Utah | Code 13-75-103 to 13-75-104 | Disclose on a clear request; prominent spoken disclosure at the start of high-risk interactions in regulated occupations; safe harbor for disclosure at the outset |
| EU | AI Act, Article 50(1) | People must be informed they are interacting with an AI system unless it is obvious from the context |
The practical conclusion for a global operator is simple. Use one opening line that says the caller is an AI assistant, keep it in every language you run, and treat consent and identity content as separate obligations that sit next to it.
United States: AI voices count as an artificial voice (FCC, February 2024)
The Federal Communications Commission adopted its Declaratory Ruling, FCC 24-17, on February 2, 2024 and released it on February 8, 2024. It says the TCPA's restrictions on the use of an "artificial or prerecorded voice" encompass current AI technologies that resemble human voices or generate call content using a prerecorded voice. The consequence, in the ruling's words, is that callers must obtain prior express consent from the called party before a call that uses such a voice, absent an emergency purpose or exemption (paragraphs 2, 5 and 9).
Two details matter for an operations team:
- It is a consent rule, not a disclosure rule. The ruling brings AI voices under the existing TCPA framework. It does not itself add a sentence the agent must speak.
- Telemarketing needs more. A footnote to the ruling says that if the robocalls introduce an advertisement or contain telemarketing, the rules require prior express written consent.
Under the rule text, calls made with an artificial voice to a cellular number need the prior express consent of the called party (47 CFR 64.1200(a)(1)(iii)). For a residential line, the exemptions in 64.1200(a)(3) have conditions. For example, a commercial-purpose call that is not an advertisement or telemarketing is exempt only if the caller makes no more than three calls in any consecutive 30-day period to the line and honours opt-out requests. See our TCPA-Konformitätsleitfaden for the consent side in full.
What every artificial-voice message must already say
Even without an AI-specific rule, 47 CFR 64.1200(b) sets content for all artificial or prerecorded voice messages:
- At the beginning of the message, state clearly the identity of the business, individual or entity responsible for initiating the call. For a business, that is the name under which it is registered to do business.
- During or after the message, state clearly a telephone number of that business or entity, not the number of the autodialer or message player. The number may not be a 900 number or one with charges above normal transmission charges.
- For telemarketing messages and messages made under certain residential exemptions, the number must let people make a do-not-call request during regular business hours, and an automated opt-out mechanism is required within two seconds of the identification.
The 2024 ruling says these identification and disclosure requirements apply to any AI technology that initiates an outbound call using an artificial or prerecorded voice (FCC 24-17, paragraph 9).
This is the first thing to check in your agent script. An AI opening that says "Hi, this is your assistant" without naming the responsible business does not meet item 1.
The FCC's August 2024 proposal for AI-call disclosure
On August 7, 2024 the FCC adopted a Notice of Proposed Rulemaking and Notice of Inquiry, FCC 24-84, released August 8, 2024, in the same docket (CG Docket No. 23-362). It proposes to define an "AI-generated call" and to require callers to disclose to consumers when they receive one. It also proposes that callers using AI-generated voice disclose at the beginning of each call, and proposes protections for positive uses of AI that help people with disabilities use the telephone network.
A proposal is not a rule. We read the NPRM itself and did not confirm whether the FCC has since adopted final rules, so treat it as proposed and check the docket before you rely on either outcome. If it is adopted in this form, an opening disclosure on every AI call would already satisfy the main idea, which is another reason to build it in now.
US states: Utah
Utah's SB 226 (2025 General Session, enrolled copy) enacted Chapter 75 of Title 13 of the Utah Code, effective 7 May 2025 in that copy. The parts that matter for voice agents:
- Disclosure on request (13-75-103(1)). A supplier using generative AI to interact with an individual in connection with a consumer transaction must disclose that the individual is interacting with generative AI and not a human, if the individual asks or otherwise prompts about it. The prompt must be a clear and unambiguous request to determine whether the interaction is with a human or with AI.
- Regulated occupations (13-75-103(2) and (3)). A person providing services in a regulated occupation must prominently disclose AI use when it is a "high-risk artificial intelligence interaction", and the disclosure must be verbal at the start of a verbal interaction. High-risk interactions include collecting sensitive personal information such as health or financial data, and giving personalised advice that could reasonably be relied on for significant decisions, such as financial, legal, medical or mental health advice.
- Safe harbor (13-75-104). There is no enforcement action under 13-75-103 if the AI clearly and conspicuously discloses, at the outset of the interaction and throughout it, that it is generative AI, is not human, or is an AI assistant.
- Penalties (13-75-105). The Division of Consumer Protection may impose an administrative fine of up to $2,500 per violation.
The statute defines generative AI as a system designed to simulate human conversation that generates non-scripted outputs similar to those created by a human, with limited or no human oversight. A tightly scripted IVR may fall outside that definition, while a conversational voice agent probably does not. That is a question for counsel. Check the current Utah Code too, because we read the enrolled bill, not a later codification.
California considered a bill on artificial-voice notices in autodialer messages (AB 2905, 2023-2024 session). The California Senate Committee on Energy, Utilities and Communications analysis of the April 25, 2024 amended version summarises it as requiring autodialing devices to notify consumers if a prerecorded message uses an artificial voice. We could not open the enacted text, so this article does not state what California requires. Check the current statute directly if you call California numbers.
EU: AI Act Article 50
Regulation (EU) 2024/1689, signed at Brussels on 13 June 2024, says it applies from 2 August 2026 (Article 113), with some chapters earlier and, after a 2026 amendment, some high-risk provisions later. Article 50(1) provides that AI systems intended to interact directly with natural persons must be designed and developed so that the people concerned are informed that they are interacting with an AI system, unless this is obvious from the point of view of a reasonably well-informed, observant and circumspect person, taking into account the circumstances and the context of use. Article 50(5) says the information must be given in a clear and distinguishable manner at the latest at the time of the first interaction.
Two points for a phone deployment:
- "Obvious from context" rarely fits a phone call. A voice agent that sounds natural, answers a business line and is not labelled gives the caller little to go on. Do not plan around the exception.
- Article 50(1) is written as a design duty on providers. Whether a business that configures a third-party agent carries its own duty depends on its role under the Act and on national law, which is a question for counsel. In practice, the business that owns the customer relationship should still make the disclosure part of the script.
Article 50(2) separately requires providers of systems that generate synthetic audio to mark outputs in a machine-readable format, subject to its conditions. Calls to EU residents also raise data protection duties. See our guides on GDPR for outbound call centers Und calling in the EU.
The Regulation was amended in 2026 by Regulation (EU) 2026/1744, the Digital Omnibus on AI, published in the Official Journal on 24 July 2026. That amendment does not change the wording of Article 50(1) or 50(5). It gives providers of systems that generate synthetic audio, and that were placed on the market before 2 August 2026, until 2 December 2026 to comply with Article 50(2), and it moves the application dates of the main high-risk provisions to 2 December 2027 and 2 August 2028. Check EUR-Lex for any later changes.
Gulf and other markets: what to check with the regulator
We did not verify AI-disclosure rules for the Gulf or other regions in this article, so we state none. What you can say about DialerBee is narrower: calling-rule packs exist for UAE (TDRA), KSA (CST), Jordan (TRC) and Bahrain (TRA) calling rules only, and the Jordan national DNC register is not integrated. Those packs cover calling rules; do not treat them as an AI-disclosure control. For any market, ask the telecom regulator and local counsel two questions: is a voice agent treated as an automated or prerecorded call, and is there a duty to tell the person it is AI?
Inbound vs outbound AI calls: where the duty differs
| Frage | Outbound AI call | Inbound AI call |
|---|---|---|
| Consent to an artificial voice | TCPA consent question applies to the call you initiate | The caller called you; the TCPA consent trigger is the call you initiate, so confirm with counsel for your case |
| Identity content | 64.1200(b) content applies to artificial-voice messages | Name the business in the greeting anyway |
| Utah disclosure on request | Applies to consumer transactions | Applies to consumer transactions |
| EU Article 50(1) | Applies to interaction with natural persons | Applies to interaction with natural persons |
The difference is mostly about consent. The disclosure duties we read are written around the interaction, not the direction of the call, so the safe default is the same: say it at the start.
A disclosure line that works on a real call
Keep it short, say who you are, say you are an AI, and offer a way to a person. Spoken lines should be under about ten seconds.
- English opening: "Hello, this is the virtual assistant from [Business name]. I'm an AI assistant, not a person. How can I help?"
- Arabic opening: مرحباً، معك المساعدة الافتراضية من [اسم الشركة]. أنا مساعدة تعمل بالذكاء الاصطناعي ولست موظفة. كيف أستطيع مساعدتك؟
- If asked "am I talking to a robot?": "Yes, I'm an AI assistant for [Business name]. I can help with this, or I can connect you to a colleague. Which would you prefer?"
- Arabic answer to the same question: نعم، أنا مساعدة تعمل بالذكاء الاصطناعي. أستطيع مساعدتك، أو أحوّلك إلى موظف إذا فضّلت.
When to repeat it: if the caller asks, if the call is transferred back to the agent after a hold, and whenever the topic moves into something sensitive such as payment details or health information. The Utah safe harbor describes disclosure at the outset and throughout, so a short reminder at a topic change is cheap. Never deny being an AI when asked sincerely, and do not give the agent a human name or backstory that suggests otherwise.
Test the Arabic line by listening to a real call, not by reading a script. DialerBee's AI voice agent has run on live calls in Arabic and English. For other languages, test on a demo first.
How to set it up in DialerBee
- Open the agent in Agent Studio and set its persona and role so the agent introduces itself as an AI assistant for your business at the start of the call.
- Add a rule for the "are you a robot?" question and the offer of a person. Our human handoff guide covers how a call passes to a person.
- Place test calls and listen to the opening and the answer to the robot question.
- After launch, review recordings and transcripts. Every AI call produces a recording and transcript that show what was actually said, which is useful if you need to show your disclosure practice. See the KI-Sprachagent page and the Leitfaden zur Compliance bei Anrufaufzeichnung for recording rules.
These are compliance-supporting controls: they help you operate a disclosure practice, and they do not make a call compliant on their own. See Compliance-Kontrollen und die AI dialer compliance guide. For US calling, start with the US region page.
Häufig gestellte Fragen
Does the FCC require an AI voice to say it is AI?
Not in a rule that is in force, as far as the sources checked for this article show. The February 2024 FCC ruling (FCC 24-17) treats AI-generated voices as an artificial voice under the TCPA, so prior express consent is needed unless an emergency or exemption applies. A separate August 2024 proposal (FCC 24-84) would require AI-call disclosure at the start of each call, but it was a proposal in the document checked. Existing rule 47 CFR 64.1200(b) already requires an artificial-voice message to state who is calling and a call-back number.
Do I need consent to call someone with an AI voice agent in the US?
Yes in the cases the ruling covers. FCC 24-17 says callers must obtain prior express consent from the called party before a call that uses an artificial voice simulated or generated through AI, absent an emergency purpose or exemption. If the call includes or introduces an advertisement or is telemarketing, the rules require prior express written consent. Your counsel should confirm which category your calls fall into.
What does Utah require an AI agent to say?
Utah Code 13-75-103, enacted by SB 226 and effective 7 May 2025 in the enrolled copy, requires a supplier using generative AI in a consumer transaction to disclose that the person is dealing with AI and not a human if the person clearly and unambiguously asks. In a regulated occupation, a high-risk AI interaction needs a prominent disclosure, spoken at the start of a verbal interaction. A safe harbor in 13-75-104 covers an AI that clearly and conspicuously discloses at the outset and throughout.
Does the EU AI Act require a phone agent to disclose it is AI?
Article 50(1) of Regulation (EU) 2024/1689 requires AI systems intended to interact directly with people to be designed so that the people are informed they are interacting with an AI system, unless that is obvious to a reasonably well-informed, observant and circumspect person in the circumstances. The Regulation says it applies from 2 August 2026, and the information must be given at the latest at the first interaction.
Is it safe to just disclose on every AI call?
Disclosing at the start is the simplest rule to apply across regions, and the Utah safe harbor is built on it. It does not replace consent where the TCPA requires consent, and it does not replace the identity and call-back number content that artificial-voice messages must carry. Treat disclosure as one control among several, and have counsel review your wording for each market.
Does DialerBee make my AI calls compliant with these laws?
No tool can do that on its own. In Agent Studio you set the agent's persona and role, place test calls to hear how it opens, and the recording and transcript of each AI call show what was actually said. Those are compliance-supporting controls. DialerBee does not provide legal advice or guarantee regulatory compliance.
Quellen
- FCC, Declaratory Ruling FCC 24-17, adopted February 2, 2024, released February 8, 2024. docs.fcc.gov, FCC 24-17 (abgerufen am 7. November 2026)
- FCC, Notice of Proposed Rulemaking and Notice of Inquiry FCC 24-84, adopted August 7, 2024, released August 8, 2024. docs.fcc.gov, FCC 24-84 (abgerufen am 7. November 2026)
- 47 CFR 64.1200, Delivery restrictions. eCFR, section 64.1200 (abgerufen am 7. November 2026)
- Regulation (EU) 2024/1689 (Artificial Intelligence Act), Articles 50 and 113. EUR-Lex, Regulation (EU) 2024/1689 (abgerufen am 7. November 2026)
- Regulation (EU) 2026/1744 (Digital Omnibus on AI), amending Regulation (EU) 2024/1689, Official Journal 24 July 2026. EUR-Lex, Regulation (EU) 2026/1744 (abgerufen am 7. November 2026)
- Utah S.B. 226, Artificial Intelligence Consumer Protection Amendments, 2025 General Session, enrolled copy. le.utah.gov, SB 226 enrolled (abgerufen am 7. November 2026)
- California Senate Committee on Energy, Utilities and Communications, analysis of AB 2905 (Low), version amended April 25, 2024, hearing date June 18, 2024. seuc.senate.ca.gov, AB 2905 analysis (abgerufen am 7. November 2026)
Weiterführende Lektüre
- TCPA-Konformitätsleitfaden
- Leitfaden zur Einhaltung der Vorschriften für KI-Dialer
- DSGVO für Outbound-Callcenter
- AI Voice Agent Human Handoff
- Agent Studio
Haftungsausschluss: This article is general information, not legal advice. DialerBee does not provide legal advice or guarantee regulatory compliance. It summarises the texts listed above as checked on November 7, 2026, and laws, rules and proposals may be amended or interpreted differently by regulators and courts. Confirm current requirements with qualified counsel in each market you call.
Weiterführende Lektüre: Who Pays When an AI Agent Gets It Wrong? · Legen Anrufer auf, wenn eine KI sagt, dass sie eine KI ist? · Arabic Voice Agent Greeting and AI Disclosure
Verwandte Artikel
Argentina No Llame Registry: Ley 26.951 Guide
9 Minuten Lesezeit
EinhaltungBrazil SAC Decree 13.135/2026: New Call Centre Rules
11 Minuten Lesezeit
BetriebHow to Measure AI Voice Agent Containment
10 Minuten Lesezeit
LeitfädenHow to Build a Knowledge Base for a Voice Agent
9 Minuten Lesezeit
Bereit, DialerBee in Aktion zu sehen?
Fordern Sie eine 15-minütige Live-Demo an oder starten Sie eine 14-tägige kostenlose Testphase – ohne Folien, ohne Verpflichtung. Das Telefonieren beginnt, sobald Ihr eigener SIP-Trunk oder Ihre TK-Anlage verbunden und freigegeben ist.
14-tägige kostenlose Testphase · kostenloses KI-Guthaben · keine Kreditkarte · 11 Sprachen · BYOC · Kontrollen zur Unterstützung der Compliance