On April 19, 2023, Seán Kirkpatrick, then director of the All-domain Anomaly Resolution Office, walked senators through a slide deck on UAP reporting trends covering data from 1996 to 2023. The briefing was cleared for open publication by the Department of Defense Office of Prepublication and Security Review on April 17, 2023. Eighteen months later, on November 19, 2024, AARO's new director, Dr. Jon Kosloski, presented a second open briefing to the same Senate subcommittee. Both presentations drew on AARO's cumulative reporting database. Both were official, cleared, and unclassified.

They do not say the same things.

The discrepancies are not minor rounding differences. They run to the heart of how AARO characterizes what it has collected — what shape UAP most commonly take, how many cases can be explained, and what the data actually shows. For an office whose stated mission is to minimize technical and intelligence surprise through rigorous scientific and intelligence tradecraft, the variance between its own slides is difficult to explain away.

The morphology problem

The April 2023 briefing, published on AARO's website under FOIA case 23-F-0922 and cross-referenced as DoD prepublication review case 23-S-0802, reports that the most frequently observed UAP shape is orb, round, or sphere, accounting for 52 percent of morphology reports. The November 2024 briefing, presented by Kosloski to the Subcommittee on Emerging Threats and Capabilities, puts the same category — relabeled as orb, round, sphere — at 31 percent.

That's not a small drift. It is a 21-point drop in the dominant category across roughly 18 additional months of data. Possible explanations exist: the 2024 briefing covers data through October 10, 2024, while the 2023 briefing covered through roughly early 2023, and a surge in a different category could mathematically compress the orb percentage. But the 2024 deck doesn't offer that explanation. It simply presents different numbers.

The 2024 briefing introduces a new leading category that wasn't dominant in 2023: lights, at 46 percent. The 2023 briefing lists lights at just 1 percent. That is not a data maturation story. That is a categorical revolution — unless the two presentations are using fundamentally different classification schemes, which would itself be a significant methodological problem that AARO hasn't publicly addressed.

The 2024 briefing introduces a new leading category that wasn't dominant in 2023: lights, at 46 percent.

A third version of the same data exists. The FOIA-released document 23-F-0922_2, also published on AARO's reading room, shows a morphology breakdown that matches neither briefing. In that version, sphere, round, or balloon shaped accounts for 42 percent, ambiguous sensor contact accounts for 23 percent, and not specified accounts for 22 percent. The 2023 Senate briefing lists ambiguous sensor contact at effectively zero percent. The 2024 briefing lists it at 5 percent. No document explains what changed or why the categories shifted.

The altitude data tells a cleaner story, and its consistency across presentations is worth noting: both the 2023 briefing and the 23-F-0922_2 FOIA document agree that the largest single altitude band for UAP reports is 20,000 feet — the briefing puts it at 32.2 percent, with a reference to Mt. Everest as a height comparison. That figure is stable. Which makes the morphology divergence harder to attribute to a simple data update.

The resolution problem

The 2024 Kosloski briefing includes something the 2023 Kirkpatrick briefing did not: a breakdown of closed case resolution outcomes. According to the November 2024 slides, 75 percent of closed UAP cases were resolved as balloons, 16 percent as unmanned aerial systems, 6 percent as birds, and 1 percent each as aircraft, satellites, and other. Those are the closed cases. AARO does not specify in the public slides what percentage of total cases are closed versus still in active archive.

That omission matters. The 2023 Middle East case — an MQ-9 observation of an apparent spherical UAP via electro-optical sensors — was explicitly labeled in the April 2023 briefing as unresolved, placed in what AARO calls an active archive, and described as showing no demonstration of enigmatic technical capabilities and no apparent threat to airborne-asset safety. The case characteristics were described as consistent with other metallic orb observations in the region. The 2024 briefing does not revisit it publicly. Whether it was ever closed, and if so how, is not stated.

The South Asia case from 2023 is more tangled. The April 2023 briefing describes an MQ-9 observation of a UAP apparently trailed by a potentially anomalous atmospheric wake, and offers two competing explanations on the same slide: first, that the visible trail appears to be cavitation, similar to those caused during propulsion; second, that the visible trail is a camera-software artifact caused by video-compression algorithms overlaying a captured image on the previous frame. Those two explanations — cavitation from propulsion, and a digital compression glitch — are not the same thing. The slide doesn't resolve which the analysts believe. It labels the case pending peer-review. Whether peer-review has since concluded, and what it found, is not reflected in any public AARO document available for this report.

The Go Fast discrepancy

The November 2024 briefing includes an analysis of the Go Fast video, the now-familiar Navy footage of an object skimming what appeared to be the ocean surface. AARO assesses with high confidence that the object's altitude was approximately 13,000 feet above sea level, not near the surface as it appeared, and that its calculated velocity was approximately 45 mph — consistent, the slides note, with winds of approximately 60 knots measured at that altitude at the time.

The November 2024 briefing states that the video was released by DoD in 2017. The Department of Defense officially released the Go Fast video in 2020. That is a factual error in an official, cleared presentation — a small one, but notable in a document arguing for analytical precision.

The same briefing acknowledges that high confidence analysis of the UAP's exact position was not possible given a lack of precise positional data from the observing Navy aircraft. That caveat sits in tension with the high confidence altitude and velocity claims that appear in the same bullet points. AARO doesn't explain how it reaches high confidence conclusions about an object's altitude and speed when it simultaneously lacks precise positional data. That tension is not resolved in the public slides.

The classified network problem

Separately from the Senate briefings, the war.gov UAP release portal — a tier-one government source — adds a different kind of complication. On May 22, 2026, the Department of War published two videos under the designation DOW-UAP-PR050 and DOW-UAP-PR051. Both were released in response to a March 6, 2026 request from eight members of the U.S. House of Representatives for access to 51 potentially UAP-related records allegedly held by the Department of War and the Intelligence Community.

AARO's published description of the materials states that a collection of responsive materials was identified on a classified network, and that many of these materials lack a substantiated chain-of-custody. The Iran formation video, designated DOW-UAP-PR050, is assessed as likely derived from an infrared sensor aboard a U.S. military platform operating within the CENTCOM area of responsibility in 2022. The Syrian acceleration video, DOW-UAP-PR051, carries the same CENTCOM-derived assessment for 2021. Both were uploaded to the classified network in June 2024 — by whom, and under what authorization, is not stated in any public document.

The Syrian video description adds a further wrinkle: the media was digitally altered prior to its upload to a classified network, and is presented as received. AARO's release does not say who altered it, when the alterations were made relative to the June 2024 upload, or what the original footage showed before modification. Those are not rhetorical gaps — they are the kind of chain-of-custody questions that AARO itself flags as a concern with this collection of materials.

The picture that emerges across these documents is of an agency presenting data publicly with more confidence than its own internal caveats support, using classification schemes that appear to shift between presentations without explanation, and releasing materials it has explicitly described as lacking verified provenance. None of that proves malfeasance. It may reflect the genuine difficulty of building a new bureaucratic infrastructure under congressional pressure, with inherited data of uneven quality.

What would clarify the picture considerably: the peer-review findings on the South Asia case, a public methodological explanation for the morphology category changes between the 2023 and 2024 Senate briefings, and any documentation identifying who uploaded the altered Syrian video to the classified network in June 2024 and under whose authority.