If my business number is showing up as “Scam Likely”, the label names the company to approach: it is T-Mobile’s label, applied by First Orion. AT&T shows “Spam Risk” or “Fraud Risk” via Hiya; Verizon and Spectrum use TNS. USTelecom’s published redress best practice targets an initial status update within 24 hours.
Everything below is triage first. The steps in the next section cost nothing, involve no vendor, and are the ones we would do before anything else. Every remediation URL in the tables further down was opened on 8 September 2026; the three that did not resolve to a working form for us are named as such rather than quietly repeated.
What to do in the next 24 hours
None of this costs money. Do it in order.
- Read the label wording exactly, and write it down. The string is the diagnosis. Call your own number from a handset on T-Mobile, one on AT&T, one on Verizon and, if you can, a cable VoIP line. “Scam Likely” points at First Orion; “Spam Risk” or “Fraud Risk” points at Hiya; “Potential Spam” points at TNS; “Robocaller” points at Nomorobo. Those mappings are published by the USTelecom Industry Traceback Group’s Call Labeling and Blocking Points of Contact directory, not inferred by us.
- Work out whether you are being labelled or blocked. They are different problems with different remedies. If calls are being blocked rather than annotated, your originating carrier is receiving a machine-readable reason: under 47 CFR § 64.1200(k)(9), a terminating provider that blocks on analytics must return SIP code 603+ on IP networks or ISUP code 21 on non-IP networks, and every provider in the path must pass it along. Ask your carrier for the SIP response codes on your failed attempts; that settles the question in one email. Getting this wrong matters: USTelecom lists submissions that “incorrectly state that a call was ‘blocked’ when the call was ‘labeled’ or vice versa” among the factors that delay a resolution.
- Find the change that preceded the label. New number range, a list you started dialling on Monday, a campaign whose attempt cap was raised, a script that produces hang-ups in the first three seconds. First Orion, writing on its own spam-tag remediation post, names the patterns its models watch as “bursting call attempts, repeat dials across short windows, and an excessive rate of very short call durations (e.g., failed connections or quick hang-ups)”. Pause that specific campaign — not your whole operation.
- Do not move to a fresh number range. This is the one action that reliably makes things worse, and there is a primary source for it. See the section on rotation below.
- Do not pay anyone yet. Blocking-error redress is free by rule, and the published industry best practice says providers should not require you to register, buy a service or create an account in order to get redress. The exact wording is in the redress section below. Paid monitoring is a legitimate and separate product; paid redress is not something you should be charged for.
Then, over the next seven days
- File with the carrier showing the label, then with the analytics engine behind it. Both tables below carry the addresses. File once per engine, not once per day — the USTelecom redress best practices list “Repeated redress submissions without material change in the underlying circumstances of the caller or calls” among the requests a provider may decline to address.
- Register the numbers you intend to keep. Free, separate from redress, and not a precondition for it: the best practices say providers should not require registration in order to process a redress submission. See the registration section below for what registration does and does not do.
- Change the pacing the label was scored on. Attempt caps, spacing between dials and the share of calls that end in a few seconds. The threshold table below says which of these matters for which pattern.
- Re-test the same handsets at the end of the week and write down what each one displays. Without a before and after on the same devices, you cannot tell a resolved label from a quiet week.
Is this actually urgent? Often, no
One label, on one number, on one carrier, that you discovered by testing rather than by watching answer rates fall, is not an emergency. It is the kind of false positive the regulator openly acknowledges the system produces. The FCC’s Report to Congress on Robocalls and Transmission of Misleading or Inaccurate Caller Identification Information (27 December 2023) puts it plainly: provider blocking “can also result in false positives and false negatives”, and it says so in a passage that opens by observing that the practices of illegal robocallers are negatively affecting legitimate callers.
It is urgent when it is systemic: every number in a range, across two or more carriers, with a measurable drop in answer rate. That is not a false positive. That is your calling pattern being scored, and no amount of form-filling fixes it. The threshold table below is where to draw that line.
Where to file: the verified remediation contact table
The USTelecom Industry Traceback Group maintains the only consolidated directory of these contacts that we could locate. Its own framing: “Providers and their analytics partners have established mechanisms for callers to seek redress to fix any inadvertent call labeling or blocking concerns, as well as to register their numbers in the first instance.” The rows below are the ones we opened and confirmed on 8 September 2026. This is what we could verify, not an exhaustive list — the ITG directory also lists Altice, Big River, Fidelity, Midco, Sinch, Telzio, Truecaller and Windstream, which we have not reproduced here.
| Carrier | Label wording published by the ITG | Who does the labelling analytics | Where to file | What they ask for |
|---|---|---|---|---|
| T-Mobile | “Scam Likely” | First Orion | callreporting.t-mobile.com — listed by the ITG; the host returned HTTP 403 to our automated check, so we confirmed the listing, not the form |
Not confirmed by us; First Orion’s caller-side path is below |
| AT&T | “Spam Risk” or “Fraud Risk” | Hiya | att.com/reviewmycalllabel — redirects to a Hiya support request form on hiyahelp.zendesk.com; that host returned HTTP 403 to our automated check, so we confirmed the redirect, not the form | Support ticket with your number and the label observed |
| Verizon | “Potential Spam” (the form also lists “Potential Fraud”, “Robo caller” and “SPAM?”) | TNS | voicespamfeedback.com | Up to 50 numbers, business name and address, your service providers, calls per month, date and time of a sample call, why the treatment is wrong |
| Spectrum (Charter), wireline | “Spam Likely” | TNS | reportspam.spectrum.com/charter — choose “I make legitimate calls to Spectrum customers and should not be blocked or marked as spam.” | Up to 20 numbers, the label seen (“Robocaller”, “Potential SPAM” or “Potential Fraud”), date and time of a sample call, contact details. Spectrum Mobile is handled separately |
| Comcast Xfinity Voice | “Blocked – High Spam Risk” or a CNAM prepend of “Spam?” | Comcast | xfinity.com/support/articles/report-call-blocking-errors, or 844-963-0215 | Per the ITG listing; page confirmed live |
| Frontier and Lumen | “Robocaller” | Nomorobo | nomorobo.com/contact and choose “Report a Number” (Frontier also lists [email protected]) | Contact form; “Report a Number” confirmed present in the category list |
| US Cellular and C Spire | “Potential Spam” | TNS | [email protected] (C Spire also lists [email protected]) | Email; TNS’s web intake is below |
The analytics engines also take submissions directly, which is where to go when the same label follows you across several carriers:
| Analytics engine | Caller-side page | What it is for | Confirmed 8 Sep 2026 |
|---|---|---|---|
| First Orion | firstorion.com/products/free-number-registration (the ITG’s calltransparency.com redirects here) |
Free number registration. First Orion says it will “communicate to major US carriers who you are and why you’re calling, leading to more accurate call labels”, and that API access at volume “supports T-Mobile registrations only” | Yes |
| Hiya | hiya.com/manageyourcallerid (redirects to Hiya support) | “Report incorrect caller ID” under the mislabelled-spam category | Yes |
| TNS | reportarobocall.com/trf | Business feedback on blocked or mislabelled numbers; up to 20 numbers with display names, plus an email verification code before submission | Yes |
| First Orion + Hiya + TNS, in one submission | freecallerregistry.com | Registration only, not redress. Up to 20 ten-digit US numbers per submission. Read the disclaimer in the next section before relying on it | Yes |
| RoboKiller | [email protected] | The ITG also lists robokiller.com/request-block-list-exception; that URL returned HTTP 404 when we checked it, and the www variant redirected to the homepage, so we have not listed it as a working form |
Email only |
| TransUnion | callfeedback.trucontact.transunion.com/home |
Listed by the ITG for TransUnion and for Sinch-labelled traffic; the host returned HTTP 403 to our automated check, so we confirmed the listing, not the form | Listing only |
Zian AI has been running outbound acquisition since 2017, which is long enough to have watched this problem from the calling side rather than the vendor side. If you want AI phone, SMS, email and WhatsApp agents from a team that reads a label as a signal about calling behaviour rather than a form to file, Apply For Partnership to our current partnership-application beta.
What the redress process actually commits to, in writing
Two documents govern this, and neither is a vendor blog. Together they are the only published timeframes we could trace to a primary source.
The rule. Under 47 CFR § 64.1200(k)(8), every terminating provider that blocks calls or uses caller-ID authentication information in delivery decisions “must provide a single point of contact, readily available on the terminating provider’s public-facing website, for receiving call blocking error complaints”. It “must resolve disputes pertaining to caller ID authentication information within a reasonable time and, at a minimum, provide a status update within 24 hours”, and it “may not impose any charge on callers for reporting, investigating, or resolving either category of complaints, so long as the complaint is made in good faith”. Note the precise scope: the 24-hour status update in the rule attaches to caller-ID authentication disputes, and the point of contact to blocking errors. We read the operative text of paragraph (k) and did not find a labelling-specific redress deadline in it. Labelling redress is governed by industry best practice, not by that rule.
The industry best practice. The USTelecom Blocking and Labeling Working Group’s Best Practices Relating to Redress Requests, adopted 15 April 2026, sets out a separate redress process for labelled calls, with these published targets:
| Request type | Time to conclusion | Status commitment | Applies to |
|---|---|---|---|
| Standard | Up to 2 business days | “VSPs will provide an initial status update to any redress request within 24 hours.” Participating providers “commit to reaching a conclusion on at least 95% of qualifying redress requests” within these windows | Both blocked calls and labelled calls, as separate processes |
| Critical (health and safety, emergency services, fraud alerts, MFA, urgent medical notices) | Up to 1 business day |
The document is explicit that the “timeframes below are best practices and not guarantees”, and it lists what pushes a request past them: slow responses to requests for more information, complex analytics, cases spanning multiple engines or networks, a large batch of numbers, and numbers that are invalid, unassigned, unallocated or on a Do-Not-Originate list.
The most useful part for anyone being sold a fix is the restrictions section. Providers and analytics engines applying blocking or labelling “should not” require or request that you “pay a fee or execute an agreement for additional services”, market other products to you in redress correspondence, use your redress submission for marketing, “require the registration of numbers submitted for redress … in order to process the redress submission”, or “require requesters to create an account or log in before, during, or after submitting a request for redress”. If a process you are being pushed through does any of those things, it is not the redress process.
The circulating “7–14 days, 85–90% recovery” numbers: where we looked
Search this problem and you will find a consistent set of figures — a soft flag clears in 7 to 14 days of rest, hard flags take 30, recovery rates of 85 to 90 per cent. We went looking for the source. We could not locate a published cool-down period or a published recovery percentage on any of: the ITG points-of-contact directory, the USTelecom redress best practices, the operative text of 47 CFR § 64.1200, First Orion’s free registration and spam-tag remediation pages, Hiya’s support pages, TNS’s caller feedback site, Verizon’s Voice Spam Feedback form, or the Free Caller Registry terms. Every instance we could trace sat on a dialler, VoIP or lead-generation vendor blog, and none of the ones we read cited an origin.
The numbers that do have a primary source are the ones in the table above: up to 2 business days standard, up to 1 business day critical, an initial status update within 24 hours, a 95 per cent conclusion commitment — all from a named, dated USTelecom document that calls them best practices rather than guarantees. If a vendor quotes you a recovery percentage, ask which document it comes from. We are publishing this section precisely because we could not answer that question ourselves.
We also have no label-recovery measurement of our own to offer, and we are not going to imply one.
What does not work: rotating out of the flag
The instinct is to burn the number and dial from a fresh range. It is also the one instinct with a regulator’s finding pointing directly the other way, and the source is the regulator, not us.
The FCC’s Report to Congress of 27 December 2023 records, at printed pages 16–17, that illegal robocallers “often attempt to evade blocking and labeling by using VoIP services such as number rotation” — and then states the consequence for everyone else in one sentence: “because many illegal and unwanted telemarketing calls utilize number rotation, provider analytics are initially more likely to filter calls from legitimate providers who begin using new numbers, and label them as spam.”
Read that again in operational terms. Rotation is not a loophole the engines have not noticed. Rotation is the behaviour they were built to detect, so a business that responds to a flag by moving to fresh numbers is adopting the exact signature the models are tuned for, on numbers that have no history to offset it. Verizon’s own feedback form lists what its analytics weigh — “calling history and patterns, call volume, number recognition, network usage/origination, known entity, and feedback from users of the Service” — and a brand-new range scores nothing on history, recognition or known entity.
First Orion, one of the engines doing the labelling, makes the same point about paid tag removal without behaviour change on its own spam-tag remediation post: it says that approach “only delivers a short-lived lift that evaporates the moment a number’s calling behavior trips the models again”. That is the analytics engine’s own published position, and it is consistent with the rotation finding.
When a number is worth rehabilitating, and when the pattern is the problem
| What you are seeing | What it most likely is | Worth filing redress? | What actually has to change |
|---|---|---|---|
| One number, one carrier, label appeared after a discrete event (new range provisioned, one unusually heavy day) | A false positive of the kind the FCC acknowledges the system produces | Yes. This is what redress is for | Nothing structural. Let the campaign resume at its previous pace |
| One number, two or more carriers, steady dialling volume | Your pattern has crossed a scoring threshold on more than one engine | Yes, but fix pacing first | Attempt caps, spacing between dials, and the short-duration rate. USTelecom is direct: the caller “should first evaluate their calling practices to assess whether adjustments should be made to calling behavior” |
| Every number in the range flagged, across carriers, answer rate visibly down | The calling pattern itself, not a mislabel | File, but expect the treatment to be found warranted | Targeting and cadence. USTelecom notes the engine “may have data that would be contrary to the caller’s assessment of their calling practices” |
| New numbers flagged within days of provisioning, repeatedly | Rotation signature. See the FCC finding above | No. Redress will not hold | Stop rotating. Rehabilitate and keep a stable set of numbers |
| High rate of calls under about five seconds, or repeat dials to the same contact inside a short window | The patterns First Orion names as red flags for its models | Only after the pattern changes | Connection quality, list hygiene, and how many times a record is dialled |
| Calls failing outright rather than being annotated | Blocking, not labelling — a different process with different rules | Yes, and it is the category the FCC rule covers | Get the SIP 603+ or ISUP 21 codes from your carrier and file a blocking error complaint |
Registration is not redress, and the registry says so itself
Registering your numbers is free, worth doing, and frequently sold as the fix. It is not the fix. The Free Caller Registry — a single submission that reaches First Orion, Hiya and TNS — states in its own terms that “the registration of phone numbers does not guarantee redress based on analysis that each call protection provider does independently. Free Caller Registry is not to be used as a replacement for caller reputation monitoring services.” It adds that “multiple submissions of the same telephone number provide no additional value or redress functionality”, and that “neither call protection providers nor their carrier partners use Registration Data to deliver Caller ID Name” — so registering will not put your business name on the screen.
There is also an eligibility limit that catches agencies and outsourced teams. The registry is for businesses that make calls on their own behalf; its terms require you to certify that “you are not a service provider, BPO, or third party registering numbers on another business’s behalf”, and direct third parties to contact the call protection providers directly, at [email protected], [email protected] and [email protected]. First Orion’s own number-registration API, by contrast, is described as “Designed for legitimate businesses and BPOs”.
Verizon’s form carries the same caution in plainer language: “Your feedback will be shared with our third party spam analytics provider. By filling out this form, you should not assume the calls will no longer be blocked or labeled.”
The structural fix: attestation and branded calling
Redress clears a label. It does not change what the recipient sees the next time an unfamiliar number rings. The structural answer in the US is caller-ID authentication plus branded calling, and we have covered that layer in depth rather than repeating it here: our guide to branded calling and Rich Call Data for AI voice agents sets out how RCD, RFC 9795 and CTIA’s Branded Calling ID actually deliver a name, logo and call reason, and what each route costs you in vetting. Our explainer on STIR/SHAKEN attestation levels and what Australia has instead covers what A, B and C attestation do and do not assert.
The honest boundary between the two: branding and attestation change presentation and provenance; they do not exempt you from analytics. First Orion states it directly — “branding your calls does not, in and of itself, prevent spam tagging” — because the spam models score call behaviour and complaint data, which sit downstream of whether your logo renders. If your pattern is the problem, a logo makes a labelled call a labelled branded call.
Pacing is the other half. How a daily target converts into a concurrency requirement, what the carriers and platforms publish as limits, and how number reputation moves with answer rates are set out in our note on scaling AI calling from 100 to 10,000 calls — and pacing is the lever that decides whether a rehabilitated number stays clean.
If you are calling Australian numbers, the remediation path is different
“Scam Likely” is a US label from a US analytics engine on a US carrier. Australia has not mandated STIR/SHAKEN, and we could not locate an Australian equivalent of the ITG directory, a Rich Call Data programme, or a public caller-side registration path with the analytics providers. The Australian framework is Industry Code C661:2022 Reducing Scam Calls and Scam SMs, registered and enforced by the ACMA, whose rules require telcos to “identify, trace and block scam calls and SMS”; the model is built around stopping bad traffic rather than certifying good callers. If an Australian carrier is disrupting your calls, the route is your own carrier’s account team, not a public form. Number type, CLI rules and overstamping limits for AU outbound are covered in our guide to getting an Australian number for an AI voice agent.
When we are not the answer
- If calls are being blocked and the carrier will not act, the escalation is the terminating provider’s single point of contact under § 64.1200(k)(8) and, failing that, the FCC — not a vendor.
- If you dial on behalf of other businesses, the Free Caller Registry excludes you by its own terms. Go to First Orion, Hiya and TNS directly at the addresses above.
- If the label is warranted — high attempt counts, high hang-up rates, cold lists — no platform, ours included, removes it. The pattern changes or the label returns.
- If the underlying question is consent or disclosure, that is a legal question about the TCPA, state AI-disclosure rules or the Do Not Call Register, and it belongs with counsel and the regulator before it belongs with a dialling vendor.
FAQs
Why is my business number showing up as “Scam Likely”?
“Scam Likely” is T-Mobile’s label, applied by its analytics partner First Orion, according to the USTelecom Industry Traceback Group’s points-of-contact directory. It reflects a score, not an accusation: Verizon’s Voice Spam Feedback form lists the inputs its engine weighs as “calling history and patterns, call volume, number recognition, network usage/origination, known entity, and feedback from users of the Service”. A new number range, a burst of attempts, or a high rate of very short calls will all move that score.
How long does it take to get a “Scam Likely” label removed?
The only timeframes we could trace to a primary source come from the USTelecom Blocking and Labeling Working Group’s Best Practices Relating to Redress Requests, adopted 15 April 2026: up to 2 business days for a standard request, up to 1 business day for a critical one, with an initial status update within 24 hours and a commitment to conclude at least 95 per cent of qualifying requests inside those windows. The document itself says these “are best practices and not guarantees”. The 7-to-14-day cool-down figures that circulate widely could not be traced by us to any carrier, analytics engine or regulator.
Should I just move to new phone numbers?
No. It is the one common response with a regulator’s finding directly against it. The FCC’s Report to Congress of 27 December 2023 records that illegal robocallers evade blocking and labelling through number rotation, and that “because many illegal and unwanted telemarketing calls utilize number rotation, provider analytics are initially more likely to filter calls from legitimate providers who begin using new numbers, and label them as spam”. Rotation is the signature the engines were built to catch, so a fresh range usually burns faster than the one you left.
Does registering with the Free Caller Registry remove the label?
No. The registry’s own terms state that “the registration of phone numbers does not guarantee redress based on analysis that each call protection provider does independently”, and that it “is not to be used as a replacement for caller reputation monitoring services”. Registration is a separate, free, worthwhile step that gives the engines a verified business record; redress is the process that reviews a specific label. USTelecom’s best practices are explicit that a provider should not require registration as a condition of processing a redress request.
Do I have to pay someone to get a spam label removed?
Not for redress itself. Under 47 CFR § 64.1200(k)(8), a terminating provider “may not impose any charge on callers for reporting, investigating, or resolving” blocking error or caller-ID authentication complaints made in good faith. USTelecom’s redress best practices go further and say providers and analytics engines should not require you to “pay a fee or execute an agreement for additional services”, nor market other products to you in redress correspondence. Paid reputation monitoring is a legitimate separate product; being charged to file is not how the process is meant to work.
Does any of this apply to Australian numbers?
No. “Scam Likely”, “Spam Risk” and “Potential Spam” are US carrier labels. Australia has not mandated STIR/SHAKEN, and we could not locate an Australian caller-side registry, branded-calling programme or redress directory equivalent to the US ones. Under Industry Code C661:2022, Australian telcos are required to identify, trace and block scam calls, so the practical route for an Australian caller whose traffic is being disrupted is their own carrier, not a public form.
The engines score calling behaviour, so number reputation is decided long before the first word of a call. If you want AI phone, SMS, email and WhatsApp agents built by a team that treats pacing and cadence as part of the system rather than someone else’s problem, Apply For Partnership to our current partnership-application beta.