There are three separate issues: existing opt-out duties, the deferred cross-category requirement, and the newly reported FCC revisions. Mixing them produces unsafe implementation advice.
1. Existing opt-out duties remain a production requirement
The accessible current FCC regulation requires reasonable revocation requests to be honoured within a reasonable time, capped at ten business days, and says callers may not prescribe an exclusive revocation method. (ecfr.gov)
Our recommendation is a stricter operational target: suppress further eligible calls promptly after a recognised request, including queued retries. Treat ten business days as an outer legal limit under the cited rule—not as an engineering objective or permission to continue a campaign unnecessarily.
2. The January waiver addressed a narrower problem
The 6 January 2026 order extended relief concerning the application of an opt-out from one message category to unrelated future robocalls and robotexts from the caller. The extension to 31 January 2027 was not a blanket suspension of consent-revocation obligations. (docs.fcc.gov)
3. The September changes need final-text and commencement checks
The FCC’s publicly circulated September draft proposed category-specific treatment of informational-message revocations and the ability to designate certain exclusive revocation methods. Orrick subsequently reported adoption on 30 September 2026, including continued broad treatment of marketing opt-outs. (docs.fcc.gov)
Verification boundary: the final FCC 26-67 attachment returned an access error during this research. We also did not independently establish its Federal Register commencement notice. Consequently, this report does not recommend implementing the reported flexibility as though it were already operative.
Required release evidence:
- The adopted order—not only the pre-meeting draft.
- The relevant Federal Register notice and effective-date calculation.
- Counsel’s interpretation of scope across marketing, informational categories, affiliates and channels.
- Versioned changes to disclosures and preference-management logic.
- Regression tests demonstrating that old opt-outs remain respected.
What should an AI voice agent opt-out test prove?
Our proposed acceptance test should include:
- The caller says “Don’t call me again” during an interruption.
- The system recognises the request without insisting on a precise phrase.
- The event is written to the correct preference record.
- Pending redials are cancelled or blocked.
- The suppression state reaches the CRM, campaign platform and managed-service provider.
- A failed write triggers an alert and prevents further automated attempts.
- A later agent or model update does not erase the preference.
The test result should identify the caller, campaign category, relevant entity, timestamp, acknowledgement and downstream state. This is a recommended evidence design, not a claim that one particular database schema is legally mandated.