PURSUE Release 07: The Black Triangle, Read Against the FBI's Own Version

The seventh PURSUE tranche went up on war.gov/UFO on Friday, October 23, 2026. It is the smallest release yet: 28 records, every one a PDF. There is no video bundle for the first time since the program began in May. The database now holds 478 files.

The coverage led with one story, and for once the coverage picked the right one. In June 2026, according to the government’s index, federal special agents at a sensitive national security site in the western United States watched a matte black triangle rise silently over a ridge, hover, and then dive into a canyon faster than they could follow it. It is one of the most recent events in the PURSUE database. It happened four months ago.

Two documents describe it. One is AARO’s memorandum of the witness debrief. The other is the FBI’s record of the same agents’ written statement. Most outlets quoted the first. We read them side by side, and they are not the same account.

Two versions of one evening

The AARO memo, DOW-UAP-D176, is three pages. It describes a “matte black triangular shape of extremely low aspect ratio” with an estimated “thirteen-to-fifteen-foot wingspan,” six to nine feet long and one to two feet thick. It tells the event in three phases: a slow vertical climb of about ten seconds, a hover of about twenty seconds, then a two to three second “terrain-conforming descent into the canyon.” It reports no sound, no exhaust, no seams, no control surfaces, and a thermal signature that was “spatially uniform” across the whole airframe. A ground survey of the canyon the next morning “located no wreckage, ground scarring, or vegetation damage.”

The FBI file, FBI-UAP-D043, is an FD-302 that reproduces the agents’ own co-signed statement “verbatim.” Put the two next to each other:

	AARO memo (D176)	Agents’ statement in the FBI 302 (D043)
Time	2023 local	”approximately 1923 hours”
Duration	about 10 s climb, 20 s hover, 2 to 3 s exit	”The total duration of the sighting was approximately three seconds”
Wingspan	13 to 15 feet	”approximately nine to fifteen feet”
Heat	”spatially uniform thermal signature"	"appeared hotter than ambient temperature with a completely consistent thermal signature”
What it looked like	featureless craft, no propulsion visible	”most resembling a delta wing drone”
Their word for it	”unidentified craft"	"the unmanned aerial system”

Some of this is ordinary. A one-hour gap in the clock can be a time zone or daylight saving slip between two offices. A wider size range in the witnesses’ own words than in an analyst’s summary is normal. But a three-second sighting and a thirty-three-second sighting are not the same event described twice, and the hover that anchors AARO’s account does not appear in the statement the agents signed. The agents’ statement also ends by calling the object an unmanned aerial system. AARO’s memo adds that both men have “extensive operational experience” working with drones, and that on that hilltop even a small quadcopter at 1.4 miles “would have been clearly audible.”

Neither document concludes anything. AARO’s memo opens with its standard warning that the witnesses’ descriptions are their interpretations “at the time of the event,” and the index says AARO considers the case worth more work because it matches the “morphological features and kinematic profiles” of other reports from the same region.

The pictures are not photographs

The FBI file runs eleven pages. Pages three to eleven are full-page colour images, and several outlets are already calling them the first images of the triangle. They are not. We checked every frame: it is one photograph of the terrain, taken in daylight, used nine times. In each copy a black shape has been drawn in at a different point, edge-on and rising, then above the ridge, then banking down into the canyon. It is a storyboard of where the agents say the object went.

FBI-UAP-D043, pages 3 to 11, in order and cropped to the same area. A reconstruction, not footage. U.S. government work, public domain.

We sequenced the nine frames into a short clip so the path reads the way the FBI laid it out. Nothing has been added or moved.

FBI-UAP-D043 reconstruction frames, sequenced by NHIAnomalous. The triangle is drawn onto a daytime photo; this is not footage of the object.

That matters for one reason. The event happened at dusk, according to AARO, at a site where the agents were already running what the memo calls “primary optical and tripod-mounted sensor systems.” Those sensors were pointed east. The object was to the west, and the terrain “precluded quick relocation.” So the best-equipped witnesses in the PURSUE database watched through binoculars and a handheld thermal monocular, and the monocular lost it when it accelerated. As with the orbs case at the same kind of site, what was released describes evidence that would settle the question without containing any.

What the index says that the file does not

The government’s own description of D176 says the memo “further records a secondary observation on the following evening” and preserves “associated analytic support material developed by” AARO. Neither is in the three pages released. The released memo describes one evening and stops. The FBI file does not mention a second night either.

This is the second time a PURSUE western U.S. case has been released with pieces that its own paperwork refers to but the public does not have. In June it was video footage that a witness said existed and AARO said was never collected. Now it is a second sighting and an analysis.

What the coverage got wrong

• “Three separate reports” and “at least three occasions.” Release 07 has two documents on the triangle, about the same agents and the same evening. With the unreleased second night, that is two occasions at most on the record we can read.
• “Up to 1,800 miles per hour.” That figure is in neither document. The memo says the agents compared the speed to “a military fighter jet” and that “no instrumented speed measurement was obtained.” It may have come from Thursday’s press briefing; we have not seen a transcript.
• “The Pentagon released a rendering” is right. “Photos” is not.

Skinwalker Ranch: the hair was raccoon and dog

Nine of the 28 records come from the Advanced Aerospace Weapon System Applications Program, the DIA effort that paid Bigelow Aerospace Advanced Space Studies (BAASS) to study Skinwalker Ranch among much else. Unlike the research papers in Release 06, most of these had not been public.

The most important one is DOW-UAP-D174, BAASS’s own report on two field trips to “the Utah ranch” in March and April 2010. The ranch is never named, but its map labels match Skinwalker’s. The report sets out goals that include using “games, puzzles, and other strategies in order to establish communication with the source/origin of reported anomalies on the ranch property.” It collects neighbour accounts of dog-like creatures walking on two legs, small people, prehistoric-looking birds, and “a maggot which measured about 8-feet long.”

DOW-UAP-D174, page 29: a witness’s drawing of the creature he says he saw. Names redacted by the government.

Then, on page 37, it does something the books about the ranch rarely do. It prints the lab result. Four hair samples from the fence line and a red-stained rock went for mitochondrial DNA sequencing. One “matched reference Procyon lotor (raccoon) samples.” The other three shared one canine sequence, which the lab says “is frequently found in domestic dogs.” BAASS’s internal email on the next page reads it plainly: the hairs look like control samples taken from the property’s own dogs, which “run at will on the property.”

DOW-UAP-D174, page 37. The only biological lab result in seven tranches, and it is a raccoon and a dog.

That is the first laboratory result on biological material anywhere in PURSUE. It is mundane, and BAASS said so in its own email.

What BAASS told the DIA it could get

The other AAWSAP files are BAASS’s management plan and its monthly reports to the DIA from 2008 and 2009. Read them as a contractor talking to its customer, because that is what they are. Several claims in them have never appeared in a government record before:

• Crash retrievals. The December 2008 plan, DOW-UAP-D168, says “It is substantiated with testimony that several ‘crashes’ of novel spacecraft have occurred outside of the United States,” and that “BAASS affiliated personnel have a list of about ten individuals who have worked in various parts or have been affiliated with the programs involving retrieval of advanced aerospace hardware or of associated biological material.” The ten are not named.
• Contact. The same plan says it is “our understanding that select communications and contact between the operators of these spacecraft and interested parties within this country started many years ago.”
• A vault. The June 2009 report, D171, describes plans for a highly secure underground vault because of “a BAASS priority to pursue and acquire recovered hardware and biological samples from the corporate world and/or from private individuals and groups located elsewhere, including in Brazil.”
• A guard’s ankle. The November 2009 report, D173, carries a ranch security officer’s account: “The cold grip was a very slender hand that reached all the way around my ankle.”
• The wall they hit. The same report ends by saying Northern Tier, “entry into classified archives,” the oral history project and contacts with “key people in the military and private sector” were all on hold because BAASS did not have the “Need to Know.”

These are claims, not findings. None of the files contains a single piece of recovered hardware, a single named retrieval witness, or a biological sample other than the ranch hairs. The last item is the most useful: it is the contractor’s own admission that the people it said it could reach were out of reach.

The files also cut BAASS down to size in places. Its headline “Total Reports: 234,364” (D169) is mostly the UFOCAT catalogue, which the same report concedes contains duplicates. Its letters show it cutting MUFON’s monthly funding after MUFON spent about $11,000 of a $74,667 allocation. And its 2010 capability statement, D175, estimates the 2004 Tic Tac’s speed at “Mach 10.7” by assuming how long “a matter of seconds” lasted, a guess presented as an analysis.

The 1950s files, and a letter nobody answered

The rest of the release is Air Force intelligence reports from 1949 to 1957, mostly from overseas. Two stand out.

DOW-UAP-D155 is a SECRET report from January 1957. A B-52 on a Strategic Air Command mission over Saudi Arabia picked up a return on its tail gunner’s radar scope that sat at 6,000 yards for five to seven minutes, smaller than the return from the second B-52 in the cell, then slid off toward the Baghdad flight identification region. The crew drew the scope for the report.

DOW-UAP-D155, page 2: the gunnery scope as the crew drew it, January 1957. Public domain.

DOW-UAP-D181 is a handwritten letter dated May 1, 1949, from Percy Wackey of Avalon, on Catalina Island, to Kenneth Arnold, the pilot whose 1947 sighting gave the world the phrase “flying saucer.” Wackey’s wife and a schoolteacher saw six objects, “not in a group, but singly.” The letter ended up in Air Force files. Someone at Air Materiel Command intelligence wrote in the margin: “MCIAXS did not acknowledge this ltr.” It was filed as “Insufficient Data.”

DOW-UAP-D181, page 3. Public domain.

Two smaller notes for the record. The 1953 report from Sardinia, D157, says “Enclosure #1 is a sketch by the professor of what he thought he saw.” The released file has no enclosure. And a Swedish airline captain who passed an object at 600 meters below his DC-3 in December 1953, D163, told the press it seemed to be “a robot of some kind.”

Colonel McCoy’s file says nothing about saucers

DOW-UAP-D180 is the 298-page personnel file of Colonel Howard M. McCoy, who ran Air Materiel Command’s technical intelligence at Wright Field from 1946 to 1949, the years of Project SIGN and Project GRUDGE. We read all of it. It does not mention flying discs, SIGN, GRUDGE, Roswell or Arnold. His 1947 efficiency report is signed by General Nathan Twining, who calls him “My Deputy for Technical Intelligence (T-2),” and notes, “So enthusiastic about his own job that at times may tend to irate others.”

Col. Howard M. McCoy, from DOW-UAP-D180. Public domain. We have not reproduced the medical, family or identification pages in the file.

The connection to UFOs is elsewhere in this release. Tucked into BAASS’s 2008 management plan as an exhibit is Twining’s September 23, 1947 memo, “AMC Opinion Concerning ‘Flying Discs,’” the one that told the Pentagon “the phenomenon reported is something real and not visionary or fictitious.” It has been public for decades. What it says about its own sources is the point: the opinion was reached at a conference including “Intelligence T-2.” That was McCoy’s office. And it was McCoy’s intelligence department, two years later, that did not answer Mr. Wackey.

The Twining memo as it appears in DOW-UAP-D168, page 43, stamped as research material supplied to BAASS. The addressee, a public figure, is redacted in this copy.

The “foreign materiel” file is about rocket debris

The release title for EOP-UAP-D002 is “Memorandum on Interagency Coordination for the Recovery of Foreign Materiel.” Expect that phrase to travel. The memo itself never uses the word. It is a 1963 note from a White House space council staffer about a meeting with NASA’s head of international programs, Arnold Frutkin, on who should recover pieces of re-entering rockets and satellites that fall abroad. Its companion file, USG-UAP-D002, is 72 pages of the policy being drafted, from October 1962 to May 1964. We found no reference to UFOs or unidentified objects in either.

EOP-UAP-D002. Public domain.

It is still a good document, just not about aliens. Frutkin’s fourth point is the honest one: “DOD and AF want to use NASA for a ‘cover’ but have no intention of letting NASA do anything but return some pieces.” In other words, the military wanted a civilian agency’s name on the recovery of Soviet hardware. NASA’s lawyers, Frutkin added, would oppose taking anything to Congress “because of the harmful debate it would touch off.”

Where seven tranches leave us

We ask two questions of every tranche, and this one gives new answers to both.

Any biological record? One, for the first time: four hairs from Utah, tested in 2010. Raccoon and dog. The contractor that collected them told the DIA it knew of programs involving “associated biological material,” and produced none.

Any photo or video from the western U.S. cases? Still none. The new western case comes with nine drawings on a photograph and an index entry for a second night and an analysis that were not released.

The triangle is the strongest witness account PURSUE has published: trained federal agents, a second observation post, thermal equipment, a next-morning ground search. It is also two accounts that disagree on how long the thing was in view, and one of them calls it a drone. Both can be true at once. That is what unresolved means.

NHIAnomalous read all 28 records in full, about 1,300 pages. Quotes are from the released PDFs; where the scan’s text layer was garbled we read the page image.