Why AIS goes dark — and why we don’t always know why

A tanker that was actively broadcasting on AIS and has stopped is one of the most-watched signals during a Hormuz crisis. It is also one of the most ambiguous. The same observation — silence — can mean a broken antenna, a spoofed position, deliberate evasion, or a vessel that simply sailed out of the bbox we’re watching.

What AIS actually broadcasts.

AIS is a VHF-frequency radio standard mandated under the IMO’s SOLAS convention. Every vessel above 300 gross tons engaged in international voyages must carry a transponder; certain other classes (passenger ships, fishing vessels above 15m in some jurisdictions) are also required. The transponder broadcasts position, course, speed, identity (MMSI), and basic vessel attributes on a 30-second to ten-minute cadence depending on movement state.

The broadcasts are picked up by other vessels, by coastal receivers, and increasingly by satellites. The data feeds that aggregate AIS — AISStream, MarineTraffic, VesselFinder, Spire, and others — combine the terrestrial and satellite streams. The free-tier coverage we use is terrestrial-anchored, which means coverage is excellent inside ~50 km of a coastline and degrades further out. The Strait of Hormuz is well within that envelope; the central Indian Ocean is not.

The four possible meanings of silence.

Equipment failure. Transponders fail, antennas break, ships have local power problems, masters power down AIS during a bridge drill or a service. None of these is a signal of intent. The fleet baseline rate of equipment-driven silence on a given day is non-zero and noisy.

GPS spoofing. The transponder is broadcasting, but the position it reports is fictitious. This has been observed regularly in the Persian Gulf since at least 2019. Iranian vessels have been associated with spoofing during ship-to-ship transfer operations; sanctioned tankers have used it to obscure loading locations. The signature is a vessel that appears to teleport, or that broadcasts a physically-impossible track. From a one-snapshot perspective, spoofed-but-broadcasting is indistinguishable from legitimately-broadcasting unless the position is physically wrong.

Deliberate evasion. The master switches off the transponder. SOLAS allows this if “the master believes that continual operation might compromise the safety or security of his ship,” which is a deliberately soft test. In practice, AIS is switched off during sanctions-evading transfers, during transit through high-piracy zones, and during politically-sensitive port calls. It is a decision the master makes, and one which carries some legal risk under the recent OFAC guidance but limited operational consequence on the open ocean.

Bbox departure. The vessel is broadcasting normally; it has just sailed out of the area we’re watching. From our terrestrial-anchored feed, the strait bounding box is roughly lat 22–28, lng 54–60. A tanker that crosses into the western Arabian Sea is not “dark” — it’s out of frame.

The filter we apply.

The homepage AIS-gap block exists to track the deliberate- evasion case while excluding the other three. The filter is intentionally strict, biased toward false-negatives:

(1) Tanker only. Cargo, fishing, military, and passenger traffic is excluded. Tankers are the sanctioned-flow vessel class.

(2) Total sightings ≥ 50. The vessel must appear in our cumulative tracker at least fifty times before going silent. This eliminates equipment-failure noise on infrequent visitors and ensures we’re only flagging operational tankers, not first-time arrivals.

(3) Last position inside the core area. Lat 24–28, lng 55–58 — a tighter window than the broader strait bbox. A vessel last seen at the western edge of our coverage may have sailed out; a vessel last seen in the strait core didn’t.

(4) Last seen 3 to 24 hours ago. Less than 3 hours and we’re inside normal AIS reporting variance; more than 24 hours and the vessel has likely transited out and the tracker hasn’t resolved that yet. Three to twenty-four hours is the empirically-tight window for deliberate silence inside the strait.

Why the count moves with the news, but slowly.

Real escalations — carrier suspensions, missile strikes, new sanctions designations — move the AIS-gap count, but on a 24–72 hour delay rather than instantly. The vessels that disappear during a tense political week are usually the ones that were already loading in the days before; the ones that load during the tense week appear in the count a day or two later. The lag is real and unavoidable on this signal type.

Conversely, a sharp drop in the count is a meaningful signal the other way. It usually means tankers that had been silent have resumed broadcasting — either because the operational picture clarified, because the cargo has cleared, or because the vessel has left the bbox. A drop without a corresponding rise on the opposite signals (transit count, port congestion at Bandar Abbas) is the most common reading of an actual de-escalation.

What we can’t see.

The biggest blind spot is ship-to-ship transfers in the Gulf of Oman, just outside our bbox. A vessel that disables AIS on the Iranian side, transits to a transfer location, offloads, and returns broadcasting under a different identity is invisible to our filter. The trade-press accounts of these operations are circumstantial; the AIS data alone cannot prove or disprove them.

The second blind spot is renames. A tanker that goes AIS-dark, repaints the funnel, and re-emerges with a new MMSI registration may not get caught by our cumulative tracker as the same hull. The IMO number, in principle, carries through, but matching IMO across registry events is inconsistent in practice.

For these two blind spots the live count is a floor, not a ceiling. The actual operational tempo of the shadow fleet is almost certainly higher than what the dashboard surfaces.

Live AIS gap detector at /#shadow-fleet. Vessel-list JSON at /api/v1/ais-gaps. Filter methodology in full at /methodology.