Every record in the PURSUE releases carries a row in the Department of War’s public manifest, and one of that manifest’s columns is Featured. It is the government’s own flag for what it wants people to look at first, and it rotates with each tranche: the June release featured ten records, the July release featured ten, and each time the flag marked only that month’s material.
For the August 7 release it marks nine records out of forty-one. Eight of those nine are digital renderings — drawings the Federal Bureau of Investigation commissioned in 2026 from witnesses’ verbal descriptions of things they saw as long ago as 2002. There are exactly eight renderings in the tranche, and every single one is featured. The ninth featured item is the U.S. Navy’s 1953 analysis of two motion picture films.
Put plainly: of everything in this release, the material the government is most actively surfacing is almost entirely the part that is not evidence of anything.
What a rendering is, in the government’s own words
The Bureau is admirably clear about what these images are. Each rendering’s official description says it is an “artistic interpretation” of a reported incident, “based upon a first-hand description” provided by a witness, and that the Bureau “prepared this digital rendering as a visual aid in 2026 in support of the PURSUE initiative.”
Nothing there is misleading, and none of this is a case of the government overselling. The descriptions are careful. The problem is what happens when a careful caption meets a compelling picture, and the picture is the thing that travels.
Each rendering is tied to a specific interview:
The 2002 Bagram triangle rendering illustrates an FD-302 recording a March 2024 interview with a former military pilot — an account given twenty-two years after the event it describes. The 2011 dark triangle rendering illustrates an interview with two former service members. The 2023 translucent triangle and the large red-lit triangle each illustrate their own interview. And four separate renderings illustrate two interviews with two people describing the same nights in the western United States in 2026.
That last group is the one worth sitting with.
When the witnesses disagreed, the drawing still had to pick
The two 2026 interviews record two people who were present together during the same events. Their accounts do not fully agree, and the Bureau wrote the disagreement down.
One witness — who had used night optical devices for more than twenty-five years — described the lights as “dash” shaped and dull red. He also volunteered that he has astigmatism. His companion described the same objects as triangular or diamond-shaped.
That is a good interview. The agent recorded the discrepancy and recorded a plausible reason for it, which is exactly what an interview record is for. Two people looked at the same lights in the same sky and came away with different shapes, and the file preserves both.
Four renderings were made from those two accounts. Whatever they depict, they cannot depict both a dash and a diamond, and a drawing does not have the option of being uncertain. This is the structural problem with illustrating testimony: a witness can say “I think it was roughly triangular, but I’m not sure, and it was dark.” An image cannot. It has edges or it does not. The rendering resolves every ambiguity the witness left open, and it resolves it in whichever direction the artist chose, and the result looks more definite than anything the witness actually claimed.
An illustration of an account carries exactly the evidentiary weight of the account it illustrates, and not one unit more. What it adds is confidence — not evidence.
Why this is worth saying about a release we broadly welcome
None of this is an argument against the PURSUE programme, which has put 375 government records into public hands in four months and has been notably candid in its caveats. The same tranche contains genuinely valuable material: the Navy’s film analysis, a 1947 review of the Scandinavian ghost rockets, two CIA files on a 1964 radar contact near Puerto Rico, and a 1963 Brazilian crashed-sphere story that three arms of government chased and disproved in twelve days.
It is also a release that improved on its predecessors in a measurable way. Across the five tranches, the share of records published with a linked companion — a report released alongside the footage it describes, an interview alongside its illustration — has risen sharply: about 3% in the May 22 release, 32% in June, 40% in July, and 80.5% in August. The government has moved from dumping isolated files towards publishing documented sets, and that is a real methodological improvement.
Which is precisely why the featured selection is worth flagging. The most-promoted items in the best-organised release so far are the drawings. Anyone arriving at the archive through the front door meets, first, eight pictures of what people remember — and those pictures will be the images that circulate, get screenshotted, and end up illustrating articles about what the government “released.”
The honest reading is small and specific. The FBI interviewed people who reported seeing things they could not identify. It wrote their accounts down accurately, including where the accounts conflicted. Then it hired an artist. The interviews are the record. The drawings are the poster.
Sources
- Department of War UFO Disclosure Portal — the PURSUE manifest, including the
Featuredcolumn - The August 7 Pentagon file release — full briefing on the fifth tranche
- AARO (All-domain Anomaly Resolution Office) — the office that designates these cases unresolved