Case No. 347·Greece

2023·Aegean / eastern Mediterranean waters (MGRS 35SKD and 35SMV), missions out of Larissa Air Base

Greece.Two circular objects skimming the sea, two days apart

On October 27 and 29, 2023, two separate missions flown out of Larissa Air Base each logged a small, circular object flying just above the surface of the sea. The first executed several ninety-degree turns at roughly 80 mph before dropping off the sensor feed; the second moved at about 30 mph heading toward land. Both operators used the same phrase to describe them —'seemingly circular, too small to make out details'— and both marked the intelligent-control box as 'no.' The two reports were declassified together in January 2026 under the same file.

Year

2023

Tier

B

Probability

20%

Category

Incident

Limited evidence: single-witness, local or without primary verification. Three independent axes: the «tier» measures the strength of the evidence; the «probability» estimates how genuinely unexplained the case is —a natural phenomenon can remain unexplained, so it does not equal «non-prosaic»—; and the partition of explanations (below) says what it most plausibly was. So a well-documented case can have a possible hoax as its most plausible cause, and a Tier B is not, for that reason, a hoax.

Primary documents · viewer

DOW-UAP-PR34 — sensor footage, Greece, October 2023

US Department of War vía DVIDS (PURSUE Release 01) · Dominio público · US Government work · open original

DOW-UAP-PR35 — sensor footage, Greece, October 2023

US Department of War vía DVIDS (PURSUE Release 01) · Dominio público · US Government work · open original

Cargando visor…

DOW-UAP-D33 — MISREP of the October 27 sighting (90-degree turns at 80 mph)

Department of War / CENTCOM (PURSUE Release 01) · Dominio público · US Government work · open original

Cargando visor…

DOW-UAP-D35 — MISREP of the October 29 sighting (30 mph heading toward land)

Department of War / CENTCOM (PURSUE Release 01) · Dominio público · US Government work · open original

Part 01

The night in question

The first sighting occurred on October 27, 2023. The aircraft had taken off from Larissa Air Base (LGLR) at 23:39Z the previous day, on a mission that also included transiting to and landing at Muwaffaq Salti (OJMS, Jordan) to relieve another platform. At 00:35:12Z, while en route to its target, the crew detected an object flying just above the surface of the water. Over the following three minutes the object executed several ninety-degree turns at an estimated kinetic velocity of 80 mph. At 00:38Z it dropped off the sensor feed.

The operator described the object as 'seemingly circular, too small to make out details.' In the structured form he recorded the maneuvers as 'sharp ninety-degree turns,' physical state as solid, means of propulsion as unknown and —significantly— the intelligent-control box as 'no.' The object's signatures were recorded as 'none': there was no detection in any band other than video. The object's response to observer actions was 'none,' the observing aircraft's heading was southwest, and the operator's assessment was 'benign.'

The second sighting occurred two days later, on October 29, on a different mission also flown out of Larissa, departing at 15:04Z on the 28th. At 08:11Z the crew detected another object likewise flying just above the ocean surface. This one moved skimming the water toward land, at an estimated kinetic velocity of 30 mph, and dropped off the feed in the same minute it was detected. The aircraft returned to Larissa at 11:05Z.

The notable coincidence is one of language: the operator on this second mission described the object with exactly the same formula —'seemingly circular, too small to make out details'— and also marked intelligent control as 'no,' signatures as 'none,' physical state solid, maneuverability 'none' and assessment 'benign.' In this second case the anomalous-characteristics box was left empty. Both reports were declassified on January 22, 2026 by the CENTCOM chief of staff under the same file, MDR 26-0019, and released to AARO.

Part 02

Why this case moved the needle

These two reports are valuable precisely for how modest they are, and for what they reveal about how the institutional UAP corpus is generated. They are the kind of record that numerically dominates military archives: small objects, no signature in sensors other than video, observed for seconds or a few minutes, which the operator himself rates as benign and which nobody pursues. Counting them as part of the phenomenon matters as much as counting the spectacular cases, because they are the base of the distribution.

The prosaic reading is strong and should be stated first. Small circular objects, at 30 and 80 mph, skimming the sea surface and in one case heading toward land, describe seabirds fairly well —the Aegean sits on a major migratory route— and also small commercial or military drones, abundant in a theater where Greek, Turkish and allied forces operate. The ninety-degree turns are anomalous for a large bird in cruise flight, but trivial for a quadcopter. That both operators marked the intelligent-control box as 'no' suggests neither perceived directed behavior.

What the pair does contribute is a weak form of cross-corroboration. Two different crews, on missions two days apart, described with the same formula objects with the same flight profile —skimming the water, small, no signature— in the same area of operations. That proves nothing about their nature, but it does indicate a recurring and locally present phenomenon rather than an isolated perception. If they are birds or drones, what is interesting is that a surveillance system designed to detect threats is formally cataloging them as unidentified, which is itself a datum about the sensor's discrimination threshold.

The posterior reflects that reading without dressing it up: more than half the mass on prosaic explanation with a misidentification subtype, a smaller share split between human platforms —classified or adversary, both plausible in the Aegean— and a fifth on indeterminate, which is what honestly remains when the object is too small to make out details and vanishes from the feed within minutes.

Part 03

What's left on paper

Documented evidence

  1. Sighting 1: Oct 27, 2023, 00:35:12Z to 00:38Z, object skimming the sea surface, 90-degree turns at ~80 mph
  2. Sighting 2: Oct 29, 2023, 08:11Z, object skimming the water toward land at ~30 mph, lost from the feed within the same minute
  3. Both missions departed Larissa Air Base (LGLR); the first with a stop at Muwaffaq Salti (OJMS)
  4. Identical description in both reports: 'seemingly circular, too small to make out details'
  5. Both forms: physical state solid, propulsion unknown, signatures 'none', intelligent control 'no', assessment 'benign'
  6. Only the first records maneuvers: 'sharp ninety-degree turns'
  7. Both declassified January 22, 2026 under the same file MDR 26-0019

Sources

  1. war.gov/UFO — DOW-UAP-D33 y D35, MISREP de los avistamientos del 27 y 29 de octubre de 2023 (PURSUE Release 01) Both documents embedded in this case's viewer.
  2. DVIDS — DOW-UAP-PR34 y PR35, grabaciones del sensor, Grecia, octubre 2023 (PURSUE Release 01) Official DoD audiovisual mirror; both videos are embedded in this case's viewer.

Location

Aegean / eastern Mediterranean waters (MGRS 35SKD and 35SMV), missions out of Larissa Air Base · 38.50°, 23.80°

Distribution of explanations

This case is classified among the model's hypotheses: the bar splits 100% by how much each explanation weighs (the uncertainty is spread across the hypotheses the case supports). Summed across the corpus they produce the comparable partition. It is a different question from the Probability above: that one estimates how likely the case is a genuinely unexplained phenomenon; this one splits which the explanation would be.

Misidentification69%
Classified human technology15%
Adversary technology10%
Non-human + state cover-up4%
Non-human, no state management2%

Modal hypothesis: Misidentification 69% · sums to 100%

Structured analytical judgment, not a calibrated frequency. Forced classification: the mass the evidence cannot assign is spread across the hypotheses the case does support.

Related cases