TCPA Nonprofit Exemption: What It Covers for AI Voice Calls to Donors

1. What does the TCPA nonprofit exemption cover for AI voice calls?

For AI voice calls made without consent, the TCPA nonprofit exemption covers one narrow thing: calls from a tax exempt nonprofit to a residential landline, limited to three calls per line in any consecutive 30 day period, with mandatory identification and opt out features. It does not cover calls to cell phones. Other TCPA provisions give nonprofits separate, different treatment, which sections 3 and 5 explain.

Three calls is the limit for relying on the no consent exemption, not an absolute cap. A nonprofit that has the called person’s prior express consent can call that line more often. Note also that the exemption turns on tax exempt status, so simply being organized as a nonprofit is not enough.

The rule lives in 47 CFR 64.1200, the Federal Communications Commission’s implementing rule for the Telephone Consumer Protection Act. Paragraph (a)(3)(iv) exempts a residential call made by or on behalf of a tax exempt nonprofit organization only if the caller makes no more than three calls within any consecutive 30 day period to that line and honors opt out requests. The three call cap took effect on July 20, 2023, according to Venable’s summary of the FCC order. Before that date, nonprofit robocalls to landlines had no numerical limit at all.

This matters now for three dated reasons. The FCC’s new rule on revoking consent takes effect on November 25, 2026. GivingTuesday falls on December 1, 2026, the start of year end campaign season. And AI voice platforms now let a small development office generate personalized “thank you and please renew” calls at a volume that used to require a call center.

This article covers outbound voice, texts, vendor campaigns and the chatbot on your donation page. It builds on our earlier pieces on AI and donor data under Delaware’s privacy law and funder rules on AI written grant applications. Those covered what goes into the model. This one covers what comes out of it and lands on a donor’s phone.

2. Why does an AI voice count as a robocall?

An AI voice counts as a robocall because the FCC ruled on February 2, 2024 that the TCPA’s restrictions on “artificial or prerecorded voice” calls cover current AI technologies that generate human voices. The Declaratory Ruling (FCC 24-17), released February 8, 2024, says such calls “require the prior express consent of the called party” unless an emergency purpose or an exemption applies. The ruling names voice cloning specifically, explaining that a cloned voice is artificial “because a person is not speaking” it.

The ruling expressly forecloses arguments that technologies purporting to provide the equivalent of a live agent fall outside the rule. The Commission tied its answer to its earlier Soundboard ruling, which held that a live agent selecting prerecorded clips during a call still makes it a prerecorded voice call. Applied to AI, a staff member supervising a voice agent does not convert the agent’s speech into a live call. If the voice on the line is generated, the TCPA’s artificial voice rules apply.

For a nonprofit, three practical consequences follow. Your executive director’s cloned voice thanking donors is an artificial voice call, even with the director’s permission. An AI agent that answers questions in real time is still an artificial voice call. And a “human in the loop” who approves scripts or listens in does not change the classification.

The ruling also restated that every artificial voice message must identify the responsible entity and provide contact information. Under paragraph (b) of the rule, the message must state the organization’s name at the beginning and a callback number during or after the message. For calls made under the nonprofit exemption, that number must let anyone make a do not call request during regular business hours.

Paragraph (b)(3) adds two opt out duties for exempt calls. On a live answer, the message must offer an automated voice or key press opt out, with brief instructions, within two seconds after the identification, and an opt out must record the number on your do not call list and end the call immediately. When the message is left on an answering machine or voicemail, it must also give a toll free number that connects directly to that automated opt out. An ordinary office number answered during business hours does not substitute for it, so test both the live answer path and the voicemail path before launch.

Note what the ruling did not do. It classified AI speech for purposes of the consent rules; it did not create a general duty to tell the person that the voice is AI. Naming your organization is required. Saying the voice is AI generated is a recommendation we return to in section 8.

3. Can nonprofits make AI voice calls to donors’ cell phones?

Nonprofits can make AI voice calls to donors’ cell phones only with each donor’s prior express consent, because the TCPA’s cell phone prohibition has no nonprofit exemption. Paragraph (a)(1)(iii) of 47 CFR 64.1200 bars any non emergency call using an artificial or prerecorded voice to a number assigned to a cellular service unless the called party has given prior express consent. The FCC said it plainly in a 2016 enforcement advisory: there is “NO blanket exemption” for nonprofits calling or texting cell phones.

The nonprofit status helps in one place only. Paragraph (a)(2) requires prior express written consent for robocalls that include an advertisement or constitute telemarketing, but lets a call made by or on behalf of a tax exempt nonprofit proceed on prior express consent alone. So a nonprofit’s consent does not have to be a signed written agreement. It still has to exist, it still has to cover the person you are calling, and the same advisory says callers claiming consent “have the burden of proof to show that they obtained such consent.”

That burden is where small nonprofits are exposed. A donor database often merges phone numbers from event registrations, old reply cards, a peer to peer fundraising platform and volunteer sign up sheets, and few of those sources record what the donor agreed to. For qualifying nonprofit calls, consent can be oral or otherwise established; the operational question is whether you can prove valid consent that covers the call. Where you cannot, the conservative policy is to exclude the number from AI calling pending review. A missing record is not itself proof that consent is absent, but it leaves you without evidence.

The financial exposure is set by statute. Under 47 U.S.C. 227(b)(3), a person who receives an unlawful call can recover actual losses or $500 for each violation, whichever is greater, and a court may increase that to three times the amount if the violation was willful or knowing. Under section 227(g), a state attorney general who has reason to believe there is a pattern or practice of violations can sue on behalf of residents for actual losses or $500 per violation, with possible trebling, subject to that section’s procedural requirements. If 1,000 calls are established to violate the consent rule, statutory damages could total $500,000 before any discretionary trebling.

A note on texts. The FCC’s revocation rule discussed in section 4 applies to robotexts as well as robocalls, and many texting platforms raise their own consent questions. Texts do not use a voice, so the AI voice ruling does not reach them directly. If your AI tool drafts personalized donor texts and sends them through a mass texting platform, have counsel confirm how that platform’s configuration fits the TCPA’s consent rules before you rely on it.

4. What changes for nonprofit robocalls on November 25, 2026?

On November 25, 2026, the FCC’s revised consent revocation rule takes effect and lets callers designate one exclusive opt out method instead of honoring opt outs made “using any reasonable method.” The final rule was adopted September 30, 2026 as FCC 26-67 and published in the Federal Register on October 26, 2026 at 91 FR 67214. It supersedes a delayed effective date for part of the same paragraph that had previously been extended to January 31, 2027. It applies to anyone relying on consent for robocalls or robotexts, nonprofits included.

The designated method can be one or more of three options: an automated voice or key press opt out on the call, a reply text using standardized words such as “stop,” or a website or telephone number the caller provides for opt outs. The caller must “clearly and conspicuously disclose the designated method to revoke consent on the call or in the text.” A text sender meets that disclosure duty by naming at least one standardized word, but must still honor all seven (“stop,” “quit,” “end,” “revoke,” “opt out,” “cancel” and “unsubscribe”).

If you do not designate an exclusive method, the old standard continues. Revised paragraph (a)(11) says callers without a designated method must process revocations made by any reasonable means. Under that paragraph, a revocation made by a method you did not list creates a rebuttable presumption that consent was revoked once the called party produces evidence of the request, and the caller can rebut it on the totality of the circumstances. A donor telling a volunteer at a gala “please stop calling me” could be enough, depending on the facts. Either way, revocations must be honored within a reasonable time, and 10 business days is the outer limit, not a default waiting period.

The rule also narrows what an opt out cancels. For informational calls, meaning calls with no advertising or telemarketing, a caller may treat a revocation as applying only to the category of call it responded to. A donor who opts out of event reminders could still receive receipt confirmations. For calls that contain an advertisement or constitute telemarketing, one revocation still cancels all such calls from that caller.

Which category a donation appeal falls into is a real question. The FCC defines telemarketing in paragraph (f)(13) as a call “for the purpose of encouraging the purchase or rental of, or investment in, property, goods, or services.” On that definition, a pure gift appeal from a tax exempt nonprofit is generally not telemarketing, while a call promoting a gala ticket, a sponsored product or a joint promotion with a company may be. Do not build a category scheme around donation calls until counsel has signed off, because the order speaks to informational calls and does not address charitable appeals by name.

One thing the new rule does not change: calls made under an exemption still need the automated opt out the rule already requires. The FCC said so directly, noting that paragraph (b)(3) continues to require an automated voice or key press opt out on exempt artificial voice calls. Your landline campaign under the nonprofit exemption therefore needs the in call opt out, which must record the number and end the call immediately, and the voicemail toll free number, regardless of what you designate elsewhere. An exclusive revocation method governs consent revocation; it does not replace the separate do not call duties in section 5.

5. Do nonprofits have to follow the Do Not Call list?

Nonprofits do not have to scrub their own calls against the National Do Not Call Registry, but they do have to keep and honor their own internal do not call list for exempt robocalls. The federal registry rule in paragraph (c)(2) applies to “telephone solicitations,” and paragraph (f)(15) excludes any call by or on behalf of a tax exempt nonprofit organization from that definition. The same exclusion does not reach the internal list duties in paragraph (d).

Paragraph (d) applies to anyone making artificial or prerecorded voice calls under the exemptions in paragraphs (a)(3)(ii) through (v), which includes the nonprofit exemption. Its minimum standards include a written do not call policy available on demand, trained personnel, recording each request when it is made and honoring it within a reasonable time not to exceed 10 business days, identifying the caller, and keeping each request for five years. Requests made through the automated in call mechanism must be recorded automatically at the moment they are made.

Two details catch small organizations. First, a do not call request applies to the entity that made the call, and the rule says it reaches affiliated entities only if the consumer reasonably would expect them to be included. A national federation and its local chapters should decide in advance how they will share requests. Second, the rule requires the donor’s prior express permission before you share or forward their do not call request to a party other than the organization on whose behalf the call was made or an affiliate.

The abandoned call limits in paragraph (a)(7), which govern predictive dialers that ring and hang up, expressly do not cover calls made by or on behalf of tax exempt nonprofits. That is relief under the FCC rule for a nonprofit’s own campaigns. A for profit telefunder calling on your behalf must separately assess the FTC’s own abandoned call rule, discussed in section 6. It is also not a license to ignore the identification rules, which apply to every artificial voice call, or the opt out rules, which apply to every call made under the exemption.

6. What changes when a fundraising vendor places the calls?

When a for profit fundraising vendor runs a covered charitable campaign for your nonprofit, the FTC’s Telemarketing Sales Rule also applies to the vendor. The FTC’s compliance guide explains that tax exempt nonprofits making their own calls are not covered by the rule, but that the USA PATRIOT Act brought charitable solicitations by for profit telemarketers within the rule’s scope. The FTC calls these vendors telefunders, and the rule’s definition of telemarketing reaches campaigns to induce a charitable contribution that involve more than one interstate call.

The prerecorded message provision is at 16 CFR 310.4(b)(1)(v). A telefunder may deliver a prerecorded message to induce a charitable contribution only to “a member of, or previous donor to,” the nonprofit on whose behalf the call is made. Each call must ring for at least 15 seconds or four rings before disconnecting, and within two seconds of the greeting must play the required disclosures followed by an automated voice or key press opt out. Note that this two second clock runs from the called person’s completed greeting, while the FCC’s two second clock in section 2 runs from the identification. According to the FTC guide, telefunders may not place cold calls that deliver prerecorded messages.

Do not assume the FCC’s 2024 ruling settles how this FTC provision treats AI. The FCC rule speaks of an “artificial or prerecorded voice”; the FTC provision speaks of a call that delivers a “prerecorded message.” A fully real time conversational agent may or may not be treated as delivering a prerecorded message, and we found no FTC statement resolving it. Ask counsel to analyze the specific implementation, and also have the vendor assess the TSR’s separate abandoned call prohibition, which the FCC’s nonprofit exception does not reach.

Telefunders also follow entity specific do not call rules but, like nonprofits, are exempt from the National Do Not Call Registry provision. Separately from the prerecorded message question, the TSR’s required disclosures for covered charitable calls include the charity’s identity and the fact that the purpose of the call is to solicit a charitable contribution. An AI voice agent run by a vendor has to deliver those disclosures promptly, and someone at your organization should be able to show that it does.

Using a vendor does not move your TCPA exposure. The FCC exemption covers calls “made by or on behalf of” a tax exempt nonprofit, so the nonprofit is part of the analysis either way, and paragraph (d)(3) puts liability for a vendor’s missed opt outs on the organization the calls were made for. Your vendor contract should require consent records for every mobile number, immediate suppression of in call opt outs and sync to your systems well inside the 10 business day outer limit, call recordings and logs retained for at least the five year do not call period, and indemnification for violations the vendor causes.

Watch for joint promotions. They require separate analysis of whose behalf the call is made on and what it promotes. The FCC expressly lets qualifying nonprofit calls rely on prior express consent even for telemarketing content, but a call also made for a commercial seller may fall outside that treatment and trigger additional FCC and FTC requirements, including written consent for mobile numbers. The two agencies also use different vocabulary: the FTC’s definition of telemarketing expressly includes charitable contribution campaigns, while the FCC definition quoted in section 4 focuses on purchases, rentals and investments.

7. Does an AI chatbot on your donation page need a disclosure?

An AI chatbot on your donation page may need a disclosure under state law, and in Utah the duty expressly reaches charitable solicitations. Utah Code 13-77-103 requires a supplier that uses generative AI to interact with an individual “in connection with a consumer transaction” to disclose that the person is interacting with generative AI and not a human, if the person makes a clear and unambiguous request to know. Even a bot that displays an AI label at the start must answer that question truthfully. Chapter 77 borrows its definitions from Utah Code 13-11-3, which says a consumer transaction includes “a charitable solicitation.” Utah separately requires a prominent upfront disclosure for certain high risk interactions in regulated occupations, which may matter for a helpline but usually not for an ordinary donation bot.

Maine does not wait for the consumer to ask, but its scope is narrower. Public Law 2025, chapter 294 enacted 10 MRSA 1500-Y, which covers a chatbot communicating by text or voice, or “any other computer technology,” used “to engage in trade and commerce with a consumer” in a way that may mislead a reasonable consumer into believing they are dealing with a human, unless the consumer is told clearly and conspicuously. A violation is a violation of the Maine Unfair Trade Practices Act. Whether a pure donation solicitation counts as “trade and commerce” under Maine’s definition is not resolved by the statutory text alone, so get Maine specific advice.

California’s older bot law has a different trigger. Business and Professions Code 17941 bars using a bot online with intent to mislead a person about its artificial identity in order to incentivize “a purchase or sale of goods or services in a commercial transaction or to influence a vote in an election.” It is not a disclosure law for every bot; liability requires intent to mislead about the bot’s artificial identity in order to knowingly deceive about the content of the communication. A gift ask is not obviously a commercial sale. A nonprofit running a ballot measure campaign through a chatbot may implicate the election prong if those deception and intent elements are met. The statute provides a simple safe harbor: disclose that it is a bot.

The new wave of companion chatbot laws is mostly not aimed at donation bots. Connecticut Public Act 26-15, for example, excludes chatbots used only for business operational purposes such as customer service that are not marketed to consumers as companions. A peer support or helpline bot run by a human services nonprofit is a different matter and deserves its own review, as we explained for California’s 2026 AI laws.

A sensible common baseline across all three states is a clear AI label at the start of every conversation, a label that stays visible, truthful answers when someone asks whether they are talking to a person, and a working route to a human. The human handoff is our recommendation, not a requirement of these three statutes. Whether a particular implementation satisfies each law depends on its presentation, scope and the interaction.

The table below sets the common nonprofit AI outreach channels side by side. It reflects the federal rules and the three state laws discussed above as of November 2026 and is a starting point for counsel, not a substitute for review of your specific campaign.

8. What do donors expect from AI outreach?

Donors expect to be told when they are dealing with AI, and a large majority object to bots presented as people. The third annual Donor Perceptions of AI study by Nathan Chappell and Cherian Koshy surveyed 1,020 U.S. donors from September 15 to 19, 2026. It found that 70.2 percent are concerned about AI bots portrayed as humans representing a charity, and 77.0 percent say transparency about AI use is very important, up from 69.4 percent in 2024.

The study’s authors report that 716 of the 1,020 respondents chose bots portrayed as humans as a concern, about twenty points ahead of any other item on the list. Familiarity with AI rose from 31.6 percent of donors describing themselves as very familiar in 2024 to 49.7 percent in 2026. Comfort did not rise with it: 21.0 percent ranked “nothing would make me comfortable” ahead of every safeguard offered.

Two caveats belong next to those numbers. The study is self funded and was fielded through an online panel, and 78.5 percent of 2026 respondents gave $500 or less. The annual samples are independent, the research is not peer reviewed, and these are stated attitudes, not observed giving behavior. Even so, donors’ concern about concealed identity supports disclosing AI use beyond the caller identification the FCC already requires. The 2024 ruling addressed consent and the classification of AI speech; it does not require telling the donor that the voice is AI.

A simple opening line closes that gap: “This is an AI generated voice message from [your organization].” That wording is our recommendation, not wording any rule prescribes, and it sits alongside the required identification rather than replacing it. The study’s authors advise disclosing at two points, which they call “at the door” and “at the gift”: a plain language web page on what AI is used for, then a sentence or two “whenever an automated system asks for money or personal information.” The Chronicle of Philanthropy’s coverage of the study reports the same advice.

9. What is still unsettled at the FCC?

Two FCC proceedings that could change the rules for nonprofit and advocacy calls are still open as of November 5, 2026. The first is a petition from Club for Growth asking for a waiver and exemption that would let noncommercial political callers place artificial or prerecorded voice calls, including AI generated voices, to wireless numbers without prior express consent. The FCC’s Consumer and Governmental Affairs Bureau sought comment in Public Notice DA 26-940, released September 4, 2026, with replies due October 19, 2026.

In our review we found no written Bureau or Commission order disposing of that petition as of this article’s date; check the docket before relying on that. Even if it were granted, it is framed around noncommercial political calls, not charitable fundraising, so a 501(c)(3) should not read it as relief for donor calls. Advocacy organizations with affiliated 501(c)(4) arms should watch it closely.

The second is a Further Notice of Proposed Rulemaking that the FCC published separately in the Federal Register with the November 25 rule. The FCC is asking whether callers should be required to offer a one step method to revoke consent to all future robocalls. If adopted, that would layer a “revoke all” option on top of the category approach described in section 4. Build your opt out handling so it can switch to an all calls stop without a system rebuild.

The FCC also delegated authority to its consumer bureau to reorganize 47 CFR 64.1200 for readability. Paragraph letters cited in this article may move. Check the current text on eCFR before you rely on a specific paragraph number in a policy document.

10. A 25 day plan before GivingTuesday

A small nonprofit can get AI outreach into a defensible position in the 25 days between November 6 and GivingTuesday on December 1, 2026. The plan below assumes one development lead, one operations or IT person and outside counsel for a short review. It is a baseline for this giving season, not a complete TCPA compliance program.

Days 1 to 5: inventory. List every tool that can place a call, send a text or chat with a donor, including features inside your CRM, peer to peer platform and phone system. For each, record whether it uses a generated or cloned voice, who the vendor is and where opt outs are captured. Check vendor release notes for AI features that may have been enabled without anyone deciding to use them.

Days 6 to 10: split the list. Separate landline numbers from mobile numbers using your carrier or vendor’s line type data. For every mobile number you plan to call with an AI voice, find the consent record. Numbers where you cannot prove consent come off the AI voice list pending review and go to live callers or mail. That is a conservative policy choice, since nonprofit consent need not be written.

Days 11 to 15: fix the call itself. Script the opening to name your organization in the first sentence. Add a callback number that reaches a person during business hours, an automated opt out with brief instructions within two seconds of the identification that records the number and ends the call, and, for voicemail, a toll free number that connects directly to the automated opt out. Test both a live answer and a voicemail drop before launch. Cap landline calls at three per number in any rolling 30 day window and have the platform enforce the cap, not a spreadsheet.

Days 16 to 20: decide on November 25. Choose whether to designate an exclusive revocation method under the new rule. If you do, add the disclosure to every call and text. If you do not, train staff and volunteers to treat a clear request to stop, wherever it is made, as a likely revocation, and give them a form to log it. Have counsel distinguish consent revocation from the separate do not call duties that apply to the campaign, because an exclusive method does not replace the mandatory in call opt out. Confirm that in call opt outs suppress immediately, that vendor opt outs reach your CRM well inside the 10 business day outer limit, and that each request is kept for five years.

Days 21 to 25: chatbot and paper. Label any donation page chatbot as AI at the start and throughout, make sure it answers truthfully when asked whether it is a person, and give a route to a human. Write a one page do not call policy and a one page AI outreach policy, approve both at the leadership level, and file the vendor contracts and consent evidence where you can find them. If a regulator, a plaintiff’s lawyer or a board member asks in January, that file is your answer.

This article is provided for general informational purposes and reflects the state of the law as of November 2026. It is not legal advice. Regulations in this area are changing quickly. Confirm current requirements and consult qualified counsel before making decisions for your organization.