Birds may appear in some of the Pentagon's declassified UFO videos
The White House UAP science advisory panel examined the military sensor videos released by the Pentagon. According to the panel, more than 8 percent of the footage likely shows birds. The method is simple: wingbeats cause a rhythmic flicker in the infrared image, and this can be measured regardless of distance.
Since May, the U.S. Department of War has released UFO/UAP records in six batches under the PURSUE program, including videos taken with the infrared cameras of military aircraft. On these, however, the data shown on the screen is usually redacted, and none of the videos gives how far the object was from the camera. Because of this, their speed is also uncertain. Without distance, a small object close to an aircraft, such as an insect, a bird or a balloon, can move across the image the same way as a more distant, larger and faster object.
This is the problem that the first study of the White House UAP science advisory panel got around. The panel is led by Harvard astrophysicist Avi Loeb. He himself writes that the Office of the Director of National Intelligence (ODNI) tasked him with setting up the panel, which currently has 20 members. The lead author of the study is Jacob Haqq-Misra.
The six points over Kuwait
The starting point of the investigation was the video designated DOW-UAP-PR135, which the Pentagon published on September 18. It was recorded by the infrared camera of a military aircraft near Kuwait, over Qaruh Island. According to the accompanying report, a U.S. Air Force crew saw six small objects in a group on July 29, 2025, which changed direction nimbly and simultaneously. Their estimated speed was 480 miles per hour, roughly 770 km/h, and their estimated altitude was 6.7 kilometers.
The researchers measured the brightness of all six points frame by frame. Each one brightened and dimmed steadily, 7.1–7.9 times per second, each at its own rhythm, not in sync with the others. Using artificial test points, they ruled out that the flicker was caused by video compression or the sensor. If the error were caused by the video itself, all six points would flicker at the same frequency and in phase. That is not what happened here.
The cause of the flicker is the wingbeat. The visible amount of heat radiated from a bird's body depends on how large a surface it turns toward the camera with its wings and at what angle. That is why the brightness fluctuates with the rhythm of the wingbeats. The same method was also run on footage from the Galileo Project's ground-based infrared cameras near Las Vegas. There, the birds flickered at rates between 3.4 and 10 times per second, which falls within the known range of bird wingbeats, and the values of the Kuwait points also fit within it.
There is also an explanation for the high speed. According to the study, the reported apparent speed of 480 miles per hour may be a consequence of parallax. This comes out if the camera is tracking a nearly stationary group flying at 60–80 percent of the aircraft's altitude. The birds may therefore have been much closer to the aircraft than the crew assumed.
The location fits this as well. In late July, the area around Qaruh Island is home mainly to nesting seabirds, terns and Socotra cormorants, and this is the peak of the nesting season.
Ten videos, 25 bird candidates
Haqq-Misra then ran the same method on all the sensor videos released so far in the six batches. He found a total of 25 bird candidates in ten videos, roughly 8 percent of the sensor videos. One of them, the footage designated PR097, likely shows a flock of four birds, and in a later part of the footage another four birds. Another video, PR141, shows infrared and optical images side by side. The object flickers at a similar rate in both wavelengths. According to Loeb, other videos likely show birds too, but these are too faint and grainy for the flicker rate to be measured reliably.
The sober take
The strength of the method is that wingbeat frequency does not depend on distance, so it can be measured even when the flight and sensor data have been redacted. This matters because the biggest shortcoming of the PURSUE videos is precisely the lack of distance. There has been a similar error before: in the Navy's 2015 “Go Fast” video, the object that seemed to be racing over the sea was in fact much higher up and was moving at roughly the speed of the wind.
The authors themselves point out the limitations. The investigation is limited by missing data, above all the camera's altitude. A definitive result would require the aircraft's position, the sensor data and unredacted video. The study was published on arXiv as a preprint and, to our knowledge, has not yet undergone peer review.
The remaining 92 percent does not mean those are mysterious either. Loeb himself asks how many of the remaining videos may show unidentified birds or human-made drones. The method simply cannot determine this if the image is too faint or the object does not flicker.