Overview
On 22 May 2026, the U.S. Department of War released the second tranche of records under the Presidential Unsealing and Reporting System for UAP Encounters (PURSUE). The release followed the first tranche of 8 May and formed part of a rolling declassification programme initiated after President Donald Trump directed federal departments and agencies to identify and release government material relating to UAP, UFOs and alleged extraterrestrial life.
The Department described PURSUE as a government-wide effort, supported by the Office of the Director of National Intelligence, to find, review, declassify and publish unresolved UAP-related records and historical documents. It also made an important qualification: records placed in the PURSUE archive are described as unresolved, meaning that the government has not made a definitive determination about the observed phenomenon. The Department explicitly notes that insufficient data may be one reason a case remains unresolved and separates the PURSUE archive from AARO's statutory reporting of cases that have subsequently been resolved. Department of War — PURSUE initiative
Release 02 was materially different from a conventional government report. It did not present a single analytical conclusion or a curated set of cases sharing one evidential standard. Instead, it added a heterogeneous collection: historical files dating from the early Cold War, intelligence-community material, Department of Energy records and a substantial body of military sensor video. The official PURSUE page lists the second release as 70.1 MB of documents and 5.6 GB of video. Contemporary Reuters reporting described the tranche as containing 222 files. Department of War — PURSUE Release 02 Reuters — second PURSUE release, 22 May 2026
The exact numerical total is less straightforward than the headline suggests because the public archive has continued to be indexed and reorganised as later tranches were added, while independent mirrors count records, media assets and derivative files differently. For the purposes of evaluating the event, the more reliable facts are the release date, the official bulk-download sizes, and the individual records that can be traced to Department of War or AARO-hosted sources.
Among the most notable material was a 116-page historical collection concerning reported green lights, discs and fireballs around Sandia and the New Mexico nuclear-weapons complex between 1948 and 1950. Release 02 also included military infrared recordings carrying uploader-defined titles such as “Syrian UAP instant acceleration,” “4 UAP Formation Iran 26 Aug 2022 over water,” “Cigar Shaped or Fast Spherical UAP clip 15 OCT 22,” and “UFOs in formation over Persian Gulf?”. AARO / DVIDS — Syrian UAP record AARO / DVIDS — Iran formation record AARO / DVIDS — cigar/spherical UAP record AARO / DVIDS — Persian Gulf formation record
These titles can make the material sound more conclusive than the accompanying government descriptions actually are. AARO repeatedly warns that many of the responsive files lack a substantiated chain of custody, that some were digitally altered before reaching the classified network on which they were later found, and that its descriptive captions should not be interpreted as an analytical judgement about a video's validity, nature or significance.
That warning is central to assessing Release 02.
The tranche increased public access to potentially useful material. It did not convert every released file into scientifically validated evidence.
Chronology
6 March 2026 — House records request.
Eight House members request potentially UAP-related records.
22 May 2026 — PURSUE Release 02.
The second tranche makes documents and media publicly accessible.
June–August 2026 — later tranches.
Subsequent releases provide further context for the rolling programme.
Rolling release
The second tranche needs to be understood within the structure of PURSUE itself.
The Department of War says the initiative followed a presidential statement on 19 February 2026 directing the Secretary of War and other departments and agencies to begin identifying and releasing government files connected with UAP, UFOs and alleged extraterrestrial life. The Department subsequently described the task as involving dozens of agencies and tens of millions of records, some existing only on paper, and said releases would occur on a rolling basis every few weeks as material was located and cleared. Department of War — PURSUE initiative
That model has an advantage over waiting for one final historical report. Documents can enter the public domain as they are discovered and reviewed rather than remaining inaccessible until every agency has completed its work. It also creates an evidential problem.
A rolling archival release is not equivalent to a controlled scientific dataset. Records arrive from different agencies, periods and information systems. Some are original government documents with clear archival provenance. Others are later copies or uploads to classified networks for which the path from original sensor to released file is incomplete. Historical correspondence, witness narratives and infrared video can therefore appear beside one another even though they require very different methods of evaluation.
Release 02 demonstrates this particularly clearly. The Sandia material is principally an archival problem: what did officials record, who investigated the reports and what conclusions were reached at the time? The military videos are primarily a sensor and provenance problem: what instrument produced the imagery, what information survived, was the file modified, and can apparent movement be translated into physical motion? Treating those evidence classes as though they carry the same meaning simply because they were released on the same day would obscure more than it clarifies.
Congressional request
AARO's metadata accompanying a large number of the Release 02 videos provides important context concerning how they entered the release.
According to AARO, on 6 March 2026 eight members of the U.S. House of Representatives requested access to 51 potentially UAP-related records allegedly held by the Department of War and Intelligence Community. AARO then identified a collection of responsive materials stored on a classified network. AARO — Presidential UAP Transparency Initiative
This is significant because it distinguishes the files from footage necessarily created and preserved through AARO's own case-management process.
For several recordings, AARO says a user uploaded the file to a classified network months or years after the apparent incident. In some cases the media had already been digitally modified before upload. The office also states that many of the materials lack a substantiated chain of custody.
A classified storage location therefore tells us something important but limited.
It shows that the file existed inside a government information system. It does not automatically establish who originally copied it from the sensor, whether the released file is generation-one media, whether explanatory context accompanied the original, or whether the uploader-defined title accurately describes what the sensor recorded.
This distinction is especially important when the title itself contains an interpretation such as instant acceleration, formation, spherical, cigar shaped or in and out of water. Those descriptions may preserve what the uploader believed the footage showed. AARO does not present the wording as a validated technical conclusion.
Provenance
Chain of custody is often discussed in relation to physical evidence, but the same principle applies to digital sensor material.
For a video to support strong quantitative conclusions, ideally there would be a traceable path from the original sensor recording to the released file. That path would establish the platform, sensor type, recording mode, time, location, frame rate, field of view, zoom state, compression history and any processing or editing applied to the imagery.
Release 02 contains examples where parts of that chain are missing.
AARO often uses language such as “likely derived from” a particular class of sensor or platform and notes when a user later uploaded the material to a classified network. This is meaningful institutional assessment, but it is weaker than having the native recording and complete mission package.
The consequences are practical.
If an object appears to cross the frame rapidly, the apparent motion could arise from object movement, sensor movement, changes in zoom, changes in line of sight, tracking behaviour or some combination of these. If the range is unknown, angular speed cannot be converted reliably into linear speed. If the file has been transcoded or edited, displayed frame timing may not preserve every property of the original acquisition.
None of this means that the underlying observation was conventional. It means that the released derivative file may not contain enough information to determine whether it was extraordinary. That distinction runs through much of Release 02.
Syrian video
The most immediately striking example is DOW-UAP-PR051, a five-minute video whose uploader-defined title is “Syrian UAP instant acceleration.”
AARO assesses that the footage was likely derived from an infrared sensor aboard a U.S. military platform operating within the U.S. Central Command area of responsibility in 2021. It also states that a user uploaded the video to a classified network in June 2024 and that the media had been digitally altered prior to upload. AARO / DVIDS — DOW-UAP-PR051
The internal edit is visible in the released version. The video contains replay sequences, threshold enhancement, inverted black-and-white values, zooming and slow-motion presentation.
The point around which the title is built occurs near the beginning of the original sequence. An area of contrast is maintained around the centre of the frame while the sensor pans. AARO's own descriptive timeline states that at approximately 20-21 seconds the sensor stops tracking the area of contrast, causing it to rapidly exit the right side of the frame. AARO / DVIDS — DOW-UAP-PR051
This wording materially changes how the clip should be approached.
The release certainly contains an apparent rapid displacement. It does not establish from the video alone that the object instantaneously accelerated through space. A change in sensor tracking is itself capable of moving a previously centred object rapidly across the image plane.
Determining physical acceleration would require enough information to separate target motion from sensor motion and to establish distance over time. No defensible Mach number can be extracted merely from the apparent frame displacement without independently established range, viewing geometry, platform motion and sensor parameters. This does not resolve the underlying object. It does mean that the uploader-defined title should not be treated as the result of a kinematic analysis. The video is valuable partly because AARO's accompanying description exposes exactly where the interpretive problem lies.
Iran video
DOW-UAP-PR050 carries the title “4 UAP Formation Iran 26 Aug 2022 over water.”
AARO assesses that it was likely derived from an infrared sensor on a U.S. military platform operating within CENTCOM in 2022. The released description says that four areas of contrast traverse the field of view during the opening seconds and that another area of contrast enters shortly afterwards. A user uploaded the media to the classified network in June 2024. AARO / DVIDS — DOW-UAP-PR050
The word formation is again descriptive rather than physically established.
Several image-plane objects moving together can indicate coordinated objects at similar range. They can also represent objects separated substantially in depth but projected into a similar angular region, distant aircraft, birds, balloons or other objects whose physical relationship cannot be inferred from a short infrared clip alone.
The available video does not publicly supply the range data necessary to reconstruct spacing or speed. The case is therefore potentially interesting because several areas of contrast appear together, but the release alone does not establish coordinated anomalous craft. A richer mission record—radar, platform telemetry, target range, environmental information and the original sensor file—would materially change the strength of the analysis.
Morphology and alteration
DOW-UAP-PR053 provides a useful example of how object morphology can become ambiguous in infrared imagery.
The uploader-defined title is “Cigar Shaped or Fast Spherical UAP clip 15 OCT 22.” AARO assesses the footage as likely deriving from an infrared sensor on a U.S. military platform in the CENTCOM area in October 2022. It says the media was digitally altered before upload and that the released 21-second file includes slower-speed replays of a brief object transit. AARO / DVIDS — DOW-UAP-PR053
The title itself acknowledges the morphological uncertainty.
A rapidly moving or defocused point source can elongate because of motion blur, sensor integration time, compression, optical effects or processing. Conversely, a genuinely elongated object can appear point-like at lower resolution.
Without a native frame sequence, point-spread information and better range data, apparent shape in a low-resolution infrared clip should be treated cautiously. This is particularly important for UAP taxonomy. A label such as cigar can easily migrate from a description of an image artefact into a claim about the physical structure of the object. The Release 02 metadata does not justify that step.
Persian Gulf footage
DOW-UAP-PR098 is considerably longer, at more than seventeen minutes, and carries the uploader-defined title “UFOs in formation over Persian Gulf?”
AARO says the video was likely derived from an infrared sensor aboard a U.S. military platform operating in the CENTCOM area and was uploaded to a classified network in October 2019. Its description records an area of contrast being tracked, zoomed and progressively resolved into multiple distinct areas of contrast, while also noting periods in which the targets become difficult to distinguish from video grain. AARO / DVIDS — DOW-UAP-PR098
Longer duration can improve an evidential record because it provides more opportunity to observe behaviour. It does not automatically resolve scale or range.
The released footage remains primarily an angular observation. If the targets are distant, even modest physical separations can project as a compact group. If range changes while the sensor zooms, apparent spacing and morphology can change without corresponding unusual manoeuvres.
The footage is therefore more useful as a record of sustained tracking than as a standalone demonstration of extraordinary formation flight. Its analytical value would increase substantially if the associated sensor metadata and platform position were released.
Sandia records
The most historically substantial document in Release 02 is DOW-UAP-D017, a 116-page collection relating to sightings and investigations around Sandia and the New Mexico nuclear complex between 1948 and 1950.
The Department's catalogue description states that the file contains 209 sightings of reported green orbs, discs and fireballs and includes records from the Armed Forces Special Weapons Project and U.S. Air Force. It also contains material concerning attempts to examine particulate residue associated with some of the reported events. Department of War — PURSUE Release 02 / Sandia record
This material deserves a different evidential treatment from the videos. The file's value lies principally in documenting how seriously parts of the early Cold War security establishment treated repeated aerial observations near sensitive nuclear facilities.
The records show a structured effort to collect reports, classify observers, record characteristics and consider physical explanations. They also connect the green-fireball investigation to figures and organisations already important in the history of early U.S. UFO research.
That does not mean the 209 entries represent 209 independently verified anomalous craft.
The original table combines observations of different appearance and quality. Historical investigators themselves used categories including green-fireball phenomena, discs or variations, and probable meteors. Many events were brief luminous observations for which modern reconstruction is limited by the surviving information.
The significance of the archive is therefore both stronger and narrower than the simple headline.
It establishes a substantial historical military and scientific investigation around a sensitive national-security environment. It does not make all of the recorded phenomena one physical class or prove that their proximity to nuclear facilities implies a particular origin or intent.
The Sandia file merits a dedicated case deep dive of its own rather than being treated merely as one item in Release 02.
Nuclear association
The location of the Sandia reports naturally draws attention because Sandia Base and the surrounding New Mexico complex were deeply connected to the U.S. nuclear-weapons programme. Historical UAP research has long noted reports around nuclear facilities, and the recurrence of such material in government archives is worthy of systematic study. The inferential problem is one of denominators and selection.
Sensitive facilities are heavily observed. They employ trained security personnel, maintain reporting systems and preserve records at rates that may exceed ordinary civilian locations. Nuclear installations can therefore produce a richer archive even if the underlying rate of unusual phenomena is no greater than elsewhere.
At the same time, repeated observations around facilities of exceptional strategic importance cannot simply be discarded as collection bias without examination. The appropriate question is comparative: were unusual observations genuinely more common around nuclear infrastructure after accounting for observer density, security surveillance and reporting practices? Release 02 by itself does not answer that question. It supplies historical data that can contribute to one.
Unresolved status
PURSUE explicitly labels its public material as unresolved. That term needs the same care that applies to earlier ODNI and AARO statistics. An unresolved case can remain unresolved for different reasons. There may be high-quality evidence that resists ordinary attribution, or there may be too little reliable information to make any confident attribution at all. The two situations are evidentially different.
AARO's chain-of-custody warnings make that difference especially important for Release 02. Some files may contain visually striking material while lacking the contextual information required to determine what the sensor was observing.
The fact that government investigators have not reached a definitive conclusion is therefore meaningful, but it is not a positive classification of the phenomenon. It should not be read as equivalent to confirmed anomalous technology.
Equally, the presence of provenance problems should not be used to claim that every file is worthless. A derivative video can still provide clues about what was seen, and its metadata can identify where additional records should be sought.
The correct evidential weight depends upon what claim is being tested.
Declassification and authentication
Release 02 illustrates another distinction that is often lost in public discussion: declassification is not the same as authentication of every interpretation attached to a record. A government can declassify a file because the information no longer requires protection even if the content is ambiguous, mistaken or unverified. AARO's release metadata makes this explicit.
For the military videos, the office does not merely publish the uploader's title as fact. It identifies the title as uploader-defined, describes what can be seen in relatively neutral language and repeatedly states that the description is not an investigative conclusion.
This is good archival practice. The government is authenticating that these are records it located and released. In several cases it is only partially authenticating the earlier history of the media itself. The distinction is particularly important for files that entered a classified network long after the apparent event. A record's presence on a secure system demonstrates government custody at that later point. It does not necessarily reconstruct the entire provenance back to capture.
Telemetry and geometry
The strongest public claims associated with UAP video usually concern performance: very high speed, instantaneous acceleration, unusual manoeuvring, transmedium movement or the absence of conventional propulsion. These are quantitative claims. A video can visually suggest them without containing enough data to prove them.
To determine speed, an analyst generally needs a physical scale or range estimate, angular displacement and accurate timing. To determine acceleration, the change in velocity must itself be measured. To distinguish target motion from camera motion, the sensor's pointing behaviour and the platform's own movement need to be reconstructed.
Release 02 does not publicly provide complete telemetry packages for its headline videos. This means that frame-by-frame image analysis can be useful for identifying edits, tracking behaviour and geometric possibilities, but it cannot manufacture missing range information. The absence of telemetry is therefore not a minor technical complaint. It defines the ceiling on what many of the public files can establish.
Digital alteration
Several Release 02 videos are marked as having been digitally altered before upload to the classified network. That wording can sound more suspicious than it necessarily is. The released examples themselves show ordinary analytical edits: slow motion, inversion, zooming, edge enhancement and replay. Such processing may have been performed by a user attempting to inspect the target more clearly. The problem is not that all alteration is illegitimate. The problem is that analytical processing can obscure the relationship between displayed pixels and the native sensor record if the original is unavailable.
Sharpening or thresholding may make an apparent outline more visually compelling without revealing new physical structure. Slow motion can help a viewer inspect a frame sequence but does not increase the temporal resolution of the source.
For strong analysis, processed media should ideally be accompanied by the untouched source. Release 02 frequently does not provide that complete pairing.
Transparency and certainty
PURSUE Release 02 is significant as a transparency event.
A substantial body of material that had previously existed on government systems became openly accessible. Researchers no longer had to rely solely on descriptions by witnesses or officials to know that particular video files existed. Historical records could be inspected directly. Government metadata exposed provenance limitations that would otherwise remain invisible.
That is a meaningful improvement. The scientific gain is more uneven.
Historical documents can often be evaluated effectively as documentary evidence because the research question concerns what officials recorded and investigated. Sensor videos used to support physical-performance claims require much more contextual data than a visually interesting clip.
This is why Release 02 can simultaneously represent better transparency and insufficient evidence for strong technical conclusions. Those positions are not contradictory.
Later releases
Release 02 was not intended to stand alone.
The Department subsequently released further tranches on 12 June, 10 July and 7 August 2026. The current PURSUE portal therefore contains a much larger multi-agency archive than existed on 22 May. Department of War — PURSUE archive
The continuation of the programme changes how the second tranche should be interpreted.
Some records that initially appeared isolated can potentially be compared with later releases. Agencies not prominent in the first tranche became more visible. Historical collections expanded. Additional AARO material and case documentation created a wider evidential environment.
At the same time, later quantity does not retroactively repair provenance deficiencies in a particular Release 02 video. A derivative file remains derivative unless the original or associated metadata are subsequently released. This is another reason to treat PURSUE as a growing archive rather than as a sequence of “disclosures” that each carry one cumulative conclusion.
Independent commentary
Scientific reaction to PURSUE has generally depended upon the distinction between interesting material and decisive evidence.
Harvard astrophysicist Avi Loeb, who has advocated systematic scientific investigation of UAP, later wrote that the first four PURSUE tranches did not contain evidence beyond a reasonable doubt for non-human-made technology. He nevertheless argued that reducing barriers to credible data is valuable because stronger material could become scientifically testable if released with sufficient context. Avi Loeb — July 2026 reflection on PURSUE
Loeb's position is a commentary source rather than an official case assessment, but it illustrates an important middle ground. A researcher can consider parts of the archive worth investigating without treating every unresolved file as extraordinary. That is particularly appropriate for Release 02.
Historical significance
PURSUE Release 02 marks a notable stage in the U.S. government's approach to UAP transparency because it exposed the public not only to more material, but also to the messiness of the underlying evidence.
Government UAP records do not all originate from one controlled investigative programme. Some are archival files maintained for decades. Some are modern sensor recordings. Some arrive through intelligence networks or congressional requests. Some retain clear provenance; others do not.
That complexity had often been hidden behind simplified public claims about “classified UAP videos.” Release 02 makes the problem more visible.
A file can be genuine government-held material without its title being a verified description of what happened. A video can remain unresolved without demonstrating extraordinary technology. A historical record can document sustained official concern without resolving the underlying phenomenon.
These distinctions are central to serious UAP research.
The second tranche also demonstrates why declassification is only one part of transparency. Releasing a video allows public inspection, but meaningful scientific analysis may still depend upon metadata that remain classified or were never preserved. If future PURSUE releases pair imagery with native files, calibration data and operational context, the scientific value of the programme could increase substantially.
As of 26 August 2026, Release 02 should therefore be regarded as a substantial archival and transparency development whose most dramatic visual material remains evidentially incomplete.
Evidence assessment
Release 02 materially expanded the public UAP archive, especially in the area of military sensor video, but the evidential quality of the material is highly uneven.
The strongest historical contribution is the Sandia collection. Its 116 pages provide primary documentary evidence of a sustained early Cold War investigation involving reports around a strategically important nuclear complex. That record is historically significant whether or not the individual sightings ultimately had conventional causes.
The modern military videos are more difficult to assess.
They are valuable because they establish that particular pieces of media existed inside government systems and because AARO has supplied contextual descriptions that were previously unavailable. Yet many of the same descriptions explicitly identify weaknesses in provenance. Some files were uploaded to classified networks years after capture. Some were digitally modified before upload. AARO often describes the original sensor source probabilistically rather than through a complete traceable chain.
These limitations are especially important where uploader titles contain strong interpretations.
The Syrian file does show an object or area of contrast leaving the frame rapidly. AARO's own description also states that the sensor stopped tracking it at that moment. The public video therefore does not establish instantaneous physical acceleration.
The Iran and Persian Gulf footage show multiple areas of contrast that may appear grouped. Without reliable range and three-dimensional position, the videos do not establish coordinated anomalous formation flight. The cigar/spherical clip records a fast brief transit, but its apparent shape cannot confidently be separated from motion, optics and processing from the released derivative alone. None of these observations makes the underlying events uninteresting. It means the public evidence remains below the level required for robust claims about extraordinary kinematics or vehicle morphology. Release 02 is therefore more consequential as an evidence-access event than as an evidential resolution of UAP itself.
It places more records in public hands, makes previously hidden provenance problems visible and allows independent analysts to identify what additional information would be required. In that sense, the release moves the discussion in a productive direction even where it does not answer the underlying cases.
The appropriate next step is not to choose between “alien technology” and “nothing unusual” from the released clips alone. It is to seek the native sensor files, platform telemetry, range data, mission context and custody records capable of discriminating between those possibilities.
Source library
Department of War — Presidential Unsealing and Reporting System for UAP Encounters (PURSUE)
The controlling official source for the rolling-release programme. It records the presidential direction, the government's description of PURSUE, the distinction between unresolved archive material and separately reported resolved cases, and the official release dates and bulk-download sizes.
Visit the official PURSUE portal
Department of War — PURSUE release archive
Current official archive for released PURSUE material across the successive 2026 tranches. Useful for placing Release 02 within the wider rolling programme.
Browse the official PURSUE archive
Reuters — “US releases second batch of government declassified UFO files,” 22 May 2026
Contemporary independent reporting on the second tranche. Reuters reported 222 files and highlighted the 116-page Sandia collection containing 209 recorded sightings. The number is retained as Reuters' launch-day count rather than treated as a stable count of present-day archive cards.
Read the Reuters report
AARO / DVIDS — DOW-UAP-PR051, “Syrian UAP instant acceleration”
Primary government-hosted release of the five-minute video and its associated AARO metadata. Particularly important because AARO identifies the title as uploader-defined, states that the media had been digitally altered before upload and describes the apparent rapid exit as occurring when the sensor stopped tracking the area of contrast.
View the official PR051 record
AARO — Presidential UAP Transparency Initiative
Official AARO interface for released military media. Provides AARO's descriptions and provenance warnings, including the statement that the files responded to a March 2026 congressional request and that many lacked a substantiated chain of custody.
View AARO's Presidential UAP Transparency Initiative
AARO / DVIDS — DOW-UAP-PR050, “4 UAP Formation Iran 26 Aug 2022 over water”
Primary government-hosted video and metadata for the short infrared sequence containing several areas of contrast. Useful for separating the uploader-defined “formation” description from what AARO itself says is visible.
View the official PR050 record
AARO / DVIDS — DOW-UAP-PR053, “Cigar Shaped or Fast Spherical UAP clip 15 OCT 22”
Primary government-hosted release documenting a brief transit and slower replays. AARO states that the media had been digitally altered before upload, making the file particularly relevant to questions of apparent morphology and processing.
View the official PR053 record
AARO / DVIDS — DOW-UAP-PR098, “UFOs in formation over Persian Gulf?”
Primary government-hosted release of a longer infrared tracking sequence. AARO's timeline records repeated zoom and contrast changes and periods where the areas of contrast become difficult to distinguish from video grain.
View the official PR098 record
Department of War — DOW-UAP-D017, Sandia Base, 1948-1950
Official PURSUE catalogue entry for the 116-page Armed Forces Special Weapons Project and U.S. Air Force collection. The Department describes the file as containing 209 reports of green orbs, discs and fireballs around the New Mexico military/nuclear complex.
Search the official PURSUE archive for the Sandia record
PURSUE://INDEX — DOW-UAP-D017 archival transcription and indexing
Independent research index reproducing the official Sandia asset with searchable page-level OCR and source links. Useful for navigating the 116-page file, while the Department of War copy remains the primary source.
Browse the indexed Sandia file
Avi Loeb — reflection on the first four PURSUE releases, July 2026
Independent scientific commentary. Loeb argues that the releases had not, by that stage, provided evidence beyond a reasonable doubt for non-human technology while supporting continued removal of barriers to scientifically useful data. Included as a position source rather than an official evaluation of Release 02.
Read Avi Loeb's commentary