The 2026 UFO files: a checked release and version comparison

A new release can contain an old research requirement, a different contract record and a later assessment of an already reported event. Those are three different additions to the public record. The useful question is not simply how many PDFs were uploaded, but what a reader can now inspect and whether the information was already available in another form.

The Department of War's notice of 18 September 2026 announces PURSUE Release 06. The official portal's selected R06 inventory showed 75 files at this check; the unfiltered view showed 450. These are interface inventory counts, not independently authenticated events or a count of everything held by government. The comparison cutoff is 17 October 2026, 13:35 UTC. Official sixth-release notice, official portal and R06 selector.

This is a selected comparison, not a page-by-page review of all 75 files. Two R06 administrative records, D110 and D111, are compared with an earlier public solicitation copy; two associated modifications provide dated amendment context. D077 from an earlier release is compared with the May witness-summary slides it cites. The official notice and inventory were inspected at the cutoff above; the additional source comparisons were completed later on 17 October. Document bodies below are explicitly identified independent-host reproductions. A byte-for-byte match with the government's downloads remains unestablished. Read how we preserve provenance for the principles behind retaining source links and uncertainty in a comparison.

Keep three dates separate

The event date, the date of an assessment and its first public availability answer different questions. A file written in 2026 may concern an earlier event. A document newly encountered by a reader may already have been publicly available elsewhere.

The June memorandum references an earlier public item dated 8 May 2026. The retrieved four-slide reproduction of that item describes four categories of reported experience; its first slide covers the orb-launching account. D077 develops an analysis of that component, not a replacement for every account on the four slides. May summary, slide 1, June memorandum, page 1.

A bounded release comparison

The source routes are the earlier solicitation PDF, D110 reproduced PDF, D111 reproduced PDF and D077 reproduced PDF. All were inspected by this cutoff. A document's old date is not proof of its first online publication date.

An unknown comparison cell is useful. It identifies exactly what needs retrieving before a headline can describe a file as newly declassified, newly analysed or merely newly hosted. Filling every cell with an assumption would make the table look more complete while making its conclusion less reliable.

D110: an old requirement is not a list of demonstrated capabilities

The earlier solicitation's objectives and D110 share the 18 July 2008 date, the future-threat horizon through 2050 and the twelve technical study areas. In the earlier copy, the objectives are an attachment; in D110 they are a standalone four-page document. This passage-level comparison is enough to reject the claim that the general study headings first appeared because the 2026 file has a new identifier.

The layouts are not identical, however. The earlier attachment includes a personnel section naming the intended contracting officer representative and project leader; that section is not reproduced in the standalone D110 text examined here. A reader should therefore preserve the two versions rather than declare them identical or assume that one universally reveals more than the other. The reason for the difference is not established by this comparison.

The competing interpretation is consequential. Headings about propulsion, materials or spatial/temporal translation can be read as a catalogue of technologies the government possessed. The document's function supports a narrower interpretation: these were subjects for proposed research and assessment. A requirement to investigate a possibility is not a demonstration that the possibility has been realised. Earlier attachment, PDF pages 15–18; D110, sections 2–3 and 11–15.

D111: the award record adds context, but its representation needs care

D111's first pages identify contract HHM402-08-C-0072, the contractor and priced work categories. That is a meaningful difference from the comparison solicitation. Yet this is a comparison of document functions, not proof that every field first became public in 2026. A different public archive might already have held the award information.

There is an additional version clue on the inspected cover: it bears a September 2008 award date but a standard-form revision footer dated February 2012. Later-period entries also appear in the schedule. The safe conclusion is not that the file is false, nor that every page is an untouched 2008 original. Its production or consolidation history needs establishing before precise claims are made about the state of the contract on one day. D111, PDF pages 1–5.

Two separately released modifications clarify part of the contract chronology. P00001, effective 2 September 2009, describes a base-period extension and an option exercise contingent on funding. P00005, effective 28 September 2010, extends services through 21 December 2010 without additional funding for that extension. The latter helps explain the December 2010 dates in D111's schedule; it does not prove exactly how D111 was assembled. These are inspected reproduced records, not conclusions adopted from an archive's summary. D112, PDF pages 2–4, D116, PDF pages 1–3.

The reproduction's PDF metadata reports creation and modification in September 2026, consistent with later digital processing. Metadata alone does not explain the earlier form revision or certify a chain of custody. A value printed on a cover, option line items, money obligated through modifications and actual expenditure are also not automatically the same figure. The programme-record article owns the wider history; this page's contribution is the version comparison, not a reconciled account of every payment.

D077: two consequential parts of a later assessment

The May slide and June memorandum retain the central reported sequence: a briefly visible orange source, smaller red lights, multiple observers and uncertainty about whether one or several orange sources were involved. The June memorandum adds a hypothesis-by-hypothesis discussion, a proposed flare explanation for part of the activity and a provisional unresolved remainder. That is a demonstrable change in inspectable content between these copies, not evidence of a new encounter. May slide 1, June memorandum, sections 3–4.

The scope matters: the earlier slides also describe other orb and kite-like experiences. D077's percentages concern the orb-launching component it analyses, not all four categories and certainly not every UAP report.

The memorandum says the observers did not collect photographs, video or other technical data during the incident. Its later hypothesis about unrecognised technology is expressly provisional and not substantiated by technical data or physical evidence. Those qualifications must remain attached to the hypothesis. Memorandum, pages 2 and 4.

Its flare discussion offers a competing explanation for part of the reported activity while retaining a component it considers unresolved. That is a more specific position than either “everything was flares” or “the government confirmed an exotic craft”. The document's own percentage estimates concern this reported activity, not the proportion of all UFO sightings with a particular origin. Page 3.

The useful addition is therefore an inspectable line of reasoning. Readers can ask which observations belong to each proposed explanation and what could change the assessment. They should not replace an exclusion-based hypothesis with a positive measurement of unfamiliar technology.

A conclusion can be challenged without dismissing the source

An agency assessment is evidence of what that agency assessed, with whatever supporting reasoning it makes public. Its authority does not remove the need to examine the premises. Nor does a disagreement with its conclusion make every statement in the document worthless.

For example, ask whether the proposed explanation fits the same observation interval as the account it is meant to explain. Check whether the relevant position and timing are measured or estimated. Establish whether the underlying material is available for someone outside the office to repeat the comparison.

Those questions are more productive than classifying the entire file as either a revelation or a debunking. Different parts of one report can have different explanations and different levels of support. Preserving that structure is essential when comparing apparent patterns across records.

A new public record is not necessarily new public information

There are several meaningful kinds of change. A first public copy may make a previously inaccessible assertion inspectable. A higher-quality scan may reveal wording hidden by poor reproduction. An additional page may change the context. A later assessment may apply a new explanation to observations already described publicly.

These are useful changes, but they should be named accurately. A new download URL does not by itself establish a new declassification decision. An extra file does not necessarily add an independent witness. A transcript and a scan can be two representations of the same source rather than two pieces of corroboration.

For D077, the comparison establishes a more developed analysis of one component than appears on the May slide. It does not settle the first-publication history of every sentence. For D110, a passage comparison establishes substantial earlier wording. For D111, the comparison establishes a different procurement document state, with some later dates supported by modifications, but not its first-ever disclosure. These are distinct findings, not three examples of an undifferentiated revelation.

Keep the release portal and archival collection distinct

NARA's UAP collection describes records transferred from federal agencies into Record Group 615. Its collection page is an archival access route. That should not be assumed to have the same boundaries, item numbering or update cycle as a departmental release portal. NARA collection.

When following a release claim, record which system supplied the item and the identifier used there. Preserve another identifier if a matching copy appears in a different repository. Before merging them in a research log, compare the actual contents and establish whether the relationship is a duplicate, an extract, an attachment or a later revision.

The disclosure history article explains the broader access context. The records-access comparison follows a NARA transfer through to actual public PDF bodies. That is a different access chain from the PURSUE inventory, even where the subject matter overlaps.

A useful next retrieval

The remaining source work is specific: obtain government download copies for binary comparison, establish D111's full production history, and trace the earliest public copies if making a first-disclosure claim. The May predecessor and two useful contract modifications are now inspectable through the linked reproductions. The printable release-delta table preserves both the findings and the unresolved fields.

Use the document-reading worksheet to distinguish the author's account of an observation from the observation's underlying evidence. The disclosure reference provides routes into related sources. The aim is a clear change record that another researcher can check, not a larger but unexplained file count.