The FCC Reopened Whether P2P Texting Is an Autodialer

What the FCC opened for comment on 4 September#

Public notice DA 26-940 came out on 4 September 2026 from the Consumer and Governmental Affairs Bureau, seeking comment on two unrelated petitions that happen to share a docket (docs.fcc.gov, read 6 September 2026).

One of them matters to anyone running outbound texting. Joe Shields filed it on 21 May 2026, and it sat with no comment window until last week.

His ask is narrow and pointed: a declaratory ruling that a P2P system using a human being only to click a send button, moving thousands of messages an hour, is an autodialer under the TCPA.

• Comment date is 5 October 2026. Reply date is 19 October 2026. The docket is CG Docket No. 02-278 (docs.fcc.gov, read 6 September 2026).
• The second petition in the same notice is from Club for Growth, filed 31 August 2026, asking to make noncommercial political calls to wireless numbers with an artificial, prerecorded or AI-generated voice without prior express consent.
• A public notice decides nothing. The Bureau is building a record, and it can grant, deny or sit on a petition afterwards.
• The petitions are separate, so the political-voice question and the texting question move on their own merits even though the dates line up.

How clicking send became a compliance product#

Peer-to-peer texting exists as a distinct product category because of one Bureau ruling. On 25 June 2020 the FCC granted the P2P Alliance petition in DA 20-670 (docs.fcc.gov, read 6 September 2026).

The ruling made two moves. Volume alone stopped being probative, so a platform is not an autodialer merely because it can text a large number of people.

The second move is the one the market sells. A platform that cannot originate a message without a person actively and affirmatively manually dialing each one is not an autodialer under that ruling.

• The TCPA defines an autodialer as equipment with the capacity to store or produce numbers using a random or sequential number generator, and to dial them (47 U.S.C. 227(a)(1)).
• The Bureau leaned on its own 2016 Enforcement Advisory, which stated that manually placed text messages are permissible without prior express consent (DA 20-670, read 6 September 2026).
• That ruling is why vendor pages describe a rep tapping through a queue rather than a campaign firing on a schedule.
• It is also why the phrase peer-to-peer shows up in contracts and acceptable use policies far more often than in engineering docs.

The sentence this petition is aimed at#

Paragraph 11 of DA 20-670 is where the load sits. The Bureau declined to rule on whether any particular platform is an autodialer, because the record lacked the facts to check how any of them worked.

What it did say was conditional. A platform escapes only if it truly requires a person to manually dial each number and transmit each message one at a time.

The same paragraph adds the harder half: the platform must lack the capacity to transmit more than one message without a human dialing each recipient's number (docs.fcc.gov, read 6 September 2026).

• The National Consumer Law Center argued the losing side of this in 2019, saying that on high-volume P2P systems the individual involvement is so vanishingly small as to be meaningless (DA 20-670, read 6 September 2026).
• The Bureau rejected that in 2020 on the reasoning that the TCPA was never meant to stop every kind of message, only autodialed and prerecorded ones.
• Shields is running the 2019 argument again with a number attached to it, and thousands of messages an hour is a fact a plaintiff can plead.
• Any platform where one click releases a batch is describing itself out of paragraph 11 already, whatever the marketing page says.

Why a ruling cannot reach as far as the petition asks#

Ten months after DA 20-670, the Supreme Court settled the definition the Bureau had been working around. Facebook v. Duguid, decided 1 April 2021, was unanimous (supreme.justia.com, read 6 September 2026).

The holding: to be an autodialer, equipment must have the capacity to store or produce a telephone number using a random or sequential number generator. Dialing from a stored list does not qualify.

Most texting platforms send to uploaded lists of known numbers. Under Duguid they sit outside the definition regardless of who clicks what, and a Bureau ruling cannot rewrite a statute the Court has construed.

• Duguid resolved a circuit split in favor of the narrow reading, which the Third, Seventh and Eleventh Circuits already held (supreme.justia.com, read 6 September 2026).
• The live fight since then has been over the word capacity: whether a system merely able to generate numbers counts even when it never does.
• So the realistic effect of a grant is not a new prohibition. It is a shift in litigation posture and in what a vendor can claim in a sales cycle.
• The realistic effect of a denial is the status quo, with a fresh 2026 citation for every platform that markets on the 2020 ruling.

Which part of the TCPA this touches#

Section 227(b) is the target. That is the autodialer and prerecorded-voice provision, and damages run $500 per message under 47 U.S.C. 227(b)(3), trebled to $1,500 for a willful or knowing violation, with no cap.

Read it next to the other 2026 development. Steidinger v. Blackstone Medical Services (7th Cir., 14 July 2026) held that a text is not a telephone call under section 227(c)(5).

That closed private Do Not Call claims over texts in Illinois, Indiana and Wisconsin and left 227(b) completely intact. Pressure moved to the section this petition is aimed at.

• State mini-TCPA statutes survive both developments and are where B2B texting exposure has been concentrating.
• A separate change already landed: the revoke-all portion of 47 CFR 64.1200(a)(10), delayed to 31 January 2027 by DA 26-12, was replaced by FCC 26-67 on 30 September 2026 in the same docket.
• Quiet-hours and post-revocation claims remain the cheapest cases for a plaintiff to bring, because neither needs discovery into how a platform is built.
• An autodialer fight, by contrast, is expensive on both sides and turns on engineering facts nobody has published.

The defense that does not depend on any of this#

Paragraph 13 of the 2020 ruling is the part almost nobody quotes. Even where an autodialer is used, the restrictions do not apply if the sender has prior express consent (docs.fcc.gov, read 6 September 2026).

So the autodialer argument is a fallback. It matters to senders who cannot show consent, and it is close to irrelevant to a team that documents it.

Consent has a scope. The Commission has long held that someone who knowingly releases a number gives permission for messages closely related to the purpose it was given for, absent instructions otherwise.

• A program with a recorded consent basis, a timestamp and a source per contact does not change shape whichever way this petition goes.
• A program relying on the shape of the sending tool rather than the permission behind it has been carrying a single point of failure since 2018.
• Scope is where consent claims usually break. A number given to book a service appointment does not obviously cover a promotion two years later.
• None of this is legal advice, and all of it is worth putting in front of counsel before October.

Three questions to ask about your own sending#

Most teams cannot answer the first question about their own stack, which is the interesting part. The compliance story is usually inherited from a vendor page rather than tested.

Ask them in this order, because the second one makes the first one moot in most cases.

Write the answers down. If the Bureau grants the petition, the first thing any counsel asks for is exactly this.

• Does your platform require one human action per recipient, or does one action release a batch? Paragraph 11 of DA 20-670 turns on that distinction.
• What is your consent basis for each contact, where is it stored, and can you produce the timestamp and source for any single record inside a minute?
• How fast does an opt-out propagate to every system that can send, and what is the longest gap in that chain?
• Which states are your contacts in, and does your sending window respect the recipient's local time rather than yours?
• If your vendor's compliance page cites the 2020 ruling, does the product actually match the description that ruling was conditioned on?

What to do before 5 October#

Restructuring anything now would be premature. Nothing has been granted, the Supreme Court's reading of the statute stands above whatever the Bureau does, and a petition can sit in a docket for two years.

Do the audit instead. Every action worth taking here is one that pays off whether the petition is granted, denied or ignored.

Then file, or watch. Comments are open to anyone through the Commission's electronic filing system, and a sales team with real operating facts is a more useful commenter than another trade association.

• Pull a consent report for your live list and check that every contact has a basis, a timestamp and a source you could show a court.
• Time your opt-out propagation end to end and fix the slowest link, which is almost always a system that receives contacts rather than the one that sends.
• Ask your vendor, in writing, how their platform maps onto paragraph 11 of DA 20-670, and keep the answer.
• Diarise 5 October 2026 for comments and 19 October 2026 for replies, and check CG Docket No. 02-278 after both.
• Fold the FCC 26-67 opt-out changes into the same plan, since they need the same suppression plumbing.

Frequently asked questions