Two phones in the same room, one alert
A tornado warning goes out. One phone on the kitchen counter shrieks and buzzes. The other, four feet away, stays silent. Both handsets are recent and both show full bars.
It is not one thing. Wireless Emergency Alerts run on a rulebook, 47 CFR part 10, that hands different decisions to different parties. The alert originator draws the polygon. The carrier decides whether to carry alerts at all. The network decides how much text survives. The handset decides whether to draw the banner and make noise. A quiet phone is a statement about one of those links, and the rule text is specific enough to name which one.
Participation is voluntary, and the carrier has to say so
The first checkpoint surprises people who assume alerting is mandatory. Part 10 describes WEA as a system "whereby Commercial Mobile Service Providers may elect to transmit Alert Messages to the public." A carrier that declines does not get to stay quiet about it. Section 10.240 obliges it to publish a fixed notice at every point of sale, including web sites, and prints the wording. The full-declination version reads: "[[CMS provider]] presently does not transmit wireless emergency alerts. Notice required by FCC Rule 47 CFR 10.240 (Commercial Mobile Alert Service)."
The partial version matters more for diagnosis. It tells subscribers that alerts "may not be available on all devices or in the entire service area, or if a subscriber is outside of the [[CMS provider]] service area." On a regional carrier, a prepaid brand, or an MVNO, that notice settles whether the silence is a defect or the advertised product. Roaming has its own line: section 10.470 requires the roamed-upon participating carrier to support alerts "to the extent the subscriber's mobile device is configured for and technically capable of receiving WEA alerts."
What the geographic rule actually promises
Section 10.450 is where the complaint about a neighbor getting the alert first belongs. The rule sets a measurable target: "A Participating CMS Provider is considered to have matched the target area when they deliver an Alert Message to 100 percent of the target area with no more than 0.1 of a mile overshoot." A tenth of a mile is roughly 528 feet. On a county-sized polygon that tolerance is invisible; on a warning polygon that clips one side of a street, it is the whole story.
The rule then supplies its own escape hatch. A carrier whose infrastructure cannot match the polygon must deliver to an area that "best approximates the specified target area on and only on those aspects of its network infrastructure that are incapable of matching the target area." The rule names three of the circumstances that may count as technical incapability: "when the target area is outside of the Participating CMS Provider's network coverage area, when mobile devices have location services disabled, and when legacy networks or devices cannot be updated to support this functionality."
So the promise is conditional by design. Polygon-accurate delivery is what the rule asks for; a best approximation is what it accepts when the network or the handset cannot do better. Both outcomes are compliant, which is why two phones on different carriers, standing together, can legitimately behave differently on the same warning.
Location services off is a fail-open condition
Here is the fact that reverses the common mental model. Turning off location services does not hide a phone from emergency alerts. It removes the filter that would have suppressed alerts aimed elsewhere.
The Commission explained the mechanism when it adopted device-based geo-fencing. In its Second Report and Order, FCC 18-4, the agency wrote: "Geo-fencing allows mobile devices to compare their current location to the target area specified by the alert originator and to display the Alert Message only if it is located within the target area." That comparison is the phone's job, and it needs a position fix to run.
Paragraph 10 of the same order states the fallback: "WEA-capable mobile devices with location services turned off (or otherwise unavailable) at the time of Alert Message receipt should display the Alert Message by default, provided they are within a Participating CMS Provider's best approximation of the target area." The phone is not pulled into alerts from three states away. It is included in whatever coarse area the carrier broadcast to, because the device can no longer prove it sits outside the polygon. The compliance date for that requirement was November 30, 2019; equipment that cannot be updated to support it falls into the legacy branch of section 10.450 instead.
The household reading: a phone with precise location on, sitting 600 feet outside the polygon, is silent because the system worked. A phone with location off that buzzed is also the system working.
Why the same warning arrives truncated on one handset
Section 10.430 reads as a single sentence with a carve-out. A participating carrier "must support transmission of an Alert Message that contains a maximum of 360 characters of alphanumeric text." If part of its infrastructure cannot do that, it must support 90 characters "on and only on those elements of its network incapable of supporting a 360 character Alert Message."
Two things follow. Truncation is a property of the path, not of your phone's screen, so swapping handsets on the same network element will not restore the missing text. And the carve-out is narrow, permitted only on elements that cannot do better, which is why one alert can arrive complete in one neighborhood and clipped in another on the same carrier. Section 10.441 adds that carriers must support alerts carrying an embedded URL or telephone number, so a link rendered as unclickable text is a presentation question, not a corrupted message.
The opt-out menu has three doors, and one class sits outside it
Section 10.400 recognizes four classes: National Alert, Imminent Threat Alert, Child Abduction Emergency/AMBER Alert, and Public Safety Message. Section 10.280(a) then limits what a subscriber may switch off, permitting carriers to offer an opt-out for exactly three of them: Child Abduction Emergency/AMBER Alert, Imminent Threat Alert, and Public Safety Message.
National Alerts have no toggle. Section 10.500(f) lists, among the functions a WEA-capable device must perform, "Presentation of alert content to the device, consistent with subscriber opt-out selections. National Alerts must always be presented." Google's Pixel documentation says the same thing in consumer language, noting that in some countries, including the United States, certain national alerts cannot be turned off. Apple phrases it as some government alerts not being switchable.
The classification rule also constrains the originator. An Imminent Threat Alert has to clear three CAP thresholds at once: Urgency of Immediate or Expected, Severity of Extreme or Severe, and Certainty of Observed or Likely, where the rule defines Likely as having "a probability of greater than 50 percent." A forecast that misses any one of the three is not eligible for this class, which is one reason a serious-sounding advisory can dominate broadcast coverage without touching a single phone.
Section 10.400(d) closes another loop: "A Public Safety Message may only be issued in connection with an Alert Message classified in paragraphs (a), (b) or (c) of this section." A standalone Public Safety Message on your screen implies the triggering alert was sent and your phone did not present it.
The signature you can time with a stopwatch
Part 10 specifies the sound and the buzz down to fractions of a second, which gives you a cheap field test. Section 10.520(a): "The audio attention signal must have a temporal pattern of one long tone of two (2) seconds, followed by two short tones of one (1) second each, with a half (0.5) second interval between each tone. The entire sequence must be repeated twice with a half (0.5) second interval between each repetition." Section 10.530(a) prescribes the identical shape for vibration.
The pitch is fixed too. For polyphonic devices the signal "must consist of the fundamental frequencies of 853 Hz and 960 Hz transmitted simultaneously"; monophonic devices use 960 Hz alone. Section 10.530 restricts the vibration cadence to part 10 alerts and allows muting.
Why this helps: a notification arriving with your ordinary ringtone, or a single long buzz, is likely an app notification rather than a WEA. Weather apps, campus systems, and neighborhood apps are not bound by section 10.520.
What became enforceable in September 2025, and what waits until March 2028
The rule text moved recently, and the dates decide which behavior to expect today. In the Seventh Report and Order published on March 18, 2025, the Commission defined "WEA-capable mobile device" for the first time as a device complying with the part 10 subpart E equipment requirements, and redefined "mobile device" as any customer equipment used to receive commercial mobile service, which reaches cellular tablets and watches. Those definitions took effect on September 15, 2025.
The same order authorized silent alerts, which let an originator suppress the audio signal, the vibration cadence, or both, with a device-side override so users who rely on vibration can force it back on. That piece carries a compliance date of March 18, 2028. The consequence for a 2026 diagnosis: an alert that lands on your screen without sound today is not the silent-alert feature. It is a device setting, a focus mode, a mute switch, or a ringer profile.
A smaller change followed on January 14, 2026, when the Commission's deletion order struck spent provisions from part 10, including section 10.11, section 10.230, section 10.260, and paragraph (c) of section 10.450, effective March 16, 2026. Behavior did not change, but older write-ups now cite section numbers that no longer resolve.
Tests are a separate subscription
Section 10.350 splits testing into layers, and only some of it is meant to reach you. The required monthly test runs from the Federal Alert Gateway to each carrier's gateway, and carriers "shall distribute an RMT within its WEA coverage area within 24 hours of receipt by the CMS Provider Gateway unless pre-empted by actual alert traffic or unable due to an unforeseen condition." Carriers may, but need not, give handsets the ability to receive those messages.
State and local tests sit behind an explicit opt-in: "Participating CMS Providers shall provide their subscribers with the option to opt in to receive State/Local WEA Tests." That is the setting labeled Test Alerts on an iPhone, off by default in the United States, and state and local tests on a Pixel. If a scheduled municipal test did not reach you, that toggle is the first place to look, and it is a different control from the alert classes in section 10.280. Section 10.350(d) separately caps public-awareness tests at "no more than two (2) WEA tests per county (or county equivalent), per calendar year that the public receives by default", and conditions them on advance outreach, a test label where technically feasible, coordination with carriers and state and local authorities, and accessible-format notice that the test is not a real warning.
Device-side checks on iPhone and Pixel
On iPhone, the controls sit at the bottom of Settings, Notifications. Apple's documentation lists Government Alerts on by default, Test Alerts off by default and offered in the United States including Puerto Rico, Earthquake Alerts and Imminent Threat Alerts on by default, and an Improve Alert Delivery option that is off by default. Apple also states that government alerts require a SIM or eSIM from a supported carrier, the handset-side reflection of carrier election.
On Pixel, the path is Settings, Notifications, Wireless Emergency Alerts, with separate switches for extreme threats, severe threats, AMBER alerts, public safety messages, and state and local tests, plus controls for vibration and for how often an unread alert re-announces itself: once, every 2 minutes, every 15 minutes, or off. A phone set to a single announcement in a noisy room produces exactly the complaint above.
One requirement is easy to forget. Section 10.500(h) requires preservation of alert messages "in a consumer-accessible format and location for at least 24 hours or until deleted by the subscriber." That stored copy is the evidence that an alert reached a handset, and it may be gone a day later.
A diagnostic order that matches the rules
Work the chain in the order the rules stack it, rather than starting with the phone. Confirm carrier participation using the section 10.240 notice on the carrier's own pages. Compare the warning polygon published by the issuing agency against where the phone actually was, remembering the 0.1 mile tolerance and the best-approximation fallback. Then check the class, since an Imminent Threat Alert that missed a CAP threshold was not eligible to reach phones at all.
Only then does the handset matter. Check whether the relevant class is switched off, whether the test toggles are off when the message was a test, whether a focus mode silenced the cadence, and whether location services were on, which narrows rather than widens what the phone shows. Read the stored copy within 24 hours if you have one.
When This Doesn't Apply
Part 10 governs Wireless Emergency Alerts on commercial mobile networks in the United States and its territories. It does not govern the many systems that look similar on a lock screen. Opt-in municipal text programs, campus notification platforms, weather app push notifications, and neighborhood apps run on their own rules and geography, and none of them are bound by the 360-character ceiling, the 0.1 mile overshoot standard, or the 853 Hz and 960 Hz signature. Diagnosing those with part 10 produces confident wrong answers.
The rules also do not describe the broadcast Emergency Alert System, which reaches radio, television, and cable rather than handsets. A warning that crawled across a television screen without reaching a phone may have been sent to one system and not the other, which is an origination decision, not a device fault.
Outside the United States, none of this transfers. Other countries set their own character limits, tone patterns, and opt-out structures, and a phone bought abroad may present a different set of toggles. Apple documents that in some regions government alerts cannot be switched off at all.
Finally, the silent-alert provisions are not yet operative for consumers, since their compliance date is March 18, 2028. Any analysis that treats a soundless alert in 2026 as a standards feature is reading the rule ahead of its schedule.
Primary sources
Rules: Electronic Code of Federal Regulations, 47 CFR Part 10 (Wireless Emergency Alerts) and Electronic Code of Federal Regulations, 47 CFR 10.450 Geographic targeting. Geo-fencing and the location-services fallback: Federal Communications Commission, Second Report and Order FCC 18-4 (PS Docket No. 15-91). The 2025 definitions and silent-alert timetable: Federal Register, Wireless Emergency Alerts; Emergency Alert System (90 FR 12462). The January 2026 deletions: Federal Register, Delete, Delete, Delete (91 FR 1404). Handset settings: Apple Support, Government, Emergency, and Enhanced Safety Alerts on iPhone and Google Pixel Phone Help, Manage Wireless Emergency Alerts.
Two earlier pieces on this site follow the same pattern of a phone setting that does not mean what its label implies: The September Bulletin Carries Two Patch Levels and Your Phone Reports Only One and Stolen Device Protection: Which Actions Wait an Hour and Which Never Do.
Comments
Post a Comment