← The Signal Report Work with me

Report 250 · Defense Tech

The drone warnings nobody wrote down

A September GAO audit of the Secret Service found that the agency met drones near its protectees in 2015, 2016 and 2021, changed nothing about how it handles civilian drones after any of them, and kept no record of why. Then, in July 2024, a shooter in Butler, Pennsylvania flew a drone over the rally grounds for about 11 minutes before he fired. Most coverage read this as a story about the past. I read it as a story about a form, and about how a rare threat hides inside a log that only records the times you acted.

The source is GAO-26-108455, Secret Service: Protection Policies Should be Consistently Updated to Better Ensure Protectee Safety, published 3 September 2026. GAO reviewed the Secret Service's protection policies and its security-incident records for fiscal years 2015 through 2025. The drone finding takes up about a page of a 61-page report, but it is the page that explains the rest.

What GAO counted

The Secret Service calls a security incident an "unusual protective event": anything that interrupts normal protective activity, such as an unlawful entry, a suspicious vehicle, a bomb threat, or someone disrupting a protectee's speech. GAO's tally for the eleven years:

  • 83 incidents with usable records, and protection policy changed after 25 of them. That leaves 58 incidents where the agency kept the status quo.
  • There were actually 84. The agency initially couldn't find the forms for four incidents, later located three, and "could not explain why the other form was missing." GAO excluded that one.
  • For the 58 no-change decisions, policy "does not require that personnel document their rationale." So the file shows that nothing changed, and not why.

Secret Service officials told GAO that some incidents needed no change because they found no operational deficiency, the mitigation worked, and nobody was harmed. That may well be the right call much of the time. The problem GAO describes is that nobody can now check.

The three drone incidents

GAO lists them in a sidebar:

  • December 2015, Hawaii. On the return from a golf outing, an individual's drone "inadvertently came into contact with" President Obama's motorcade.
  • May 2016, San Jose. Someone flew a drone about 200 feet over the crowd at a rally for then-candidate Bernie Sanders.
  • January 2021, the inauguration. Agents saw multiple drones shortly before President Biden arrived and did not immediately recognize them as authorized U.S. Customs and Border Protection aircraft. This one did produce an immediate change: a clarification of who inside the Secret Service leads on drone countermeasures.

In none of the three did the agency update policy on what to do when a civilian puts a drone over a protectee, and it documented no rationale for leaving that alone. GAO then states the Butler facts plainly: the shooter "successfully flew the drone for approximately 11 minutes over the crowd, which helped the shooter position himself to fire."

"Some of the missing information has been relevant to subsequent attacks."

That's GAO's own sentence, from its conclusions. Its claim is careful. It does not say a written rationale would have prevented Butler. It says the rationale, had it existed, would have been available in the agency's annual review, "potentially illustrating the emerging trend of drone involvement in security incidents."

Why a blank box hides a trend

Here is the part I think most coverage missed. Three drone events in eleven years, inside a log of 83 incidents most commonly about unlawful entries, suspicious vehicles and disruptions, is a faint signal. Each event alone looked benign: an accidental bump, a hobbyist over a crowd, friendly aircraft nobody had told the detail about. If each review ends with "no deficiency, no change" and no reasoning, the annual review sees three closed files. It can't see that three different teams each decided, for their own local reasons, that drones weren't a problem yet.

A written rationale is what turns isolated judgments into data. "No change, because the drone was a hobbyist and we had no authority to act anyway" is a sentence that, read three times across six years, starts to say something. The 2021 incident is a good example of a second, different signal: the detail couldn't tell friendly drones from unknown ones. That's an identification problem, and it's the same problem every counter-drone operator still has. I wrote about the legal half of it for military bases in why some bases can't stop drones overhead, and for police in what police need before they can stop a drone.

This is the part of the report I know from the inside. In Army counter-IED work, the incident that didn't hurt anyone was still worth writing up, because the next patrol's brief was built out of those write-ups. The near miss and the "nothing happened" are where you find a new tactic before it finds you. What GAO describes is a system that recorded the decision but threw away the reasoning, and the reasoning is the part a trend lives in.

What has changed, and what hasn't

  • After Butler, a July 2025 GAO report made eight recommendations, including documenting training requirements for Secret Service counter-drone operators and documenting requirements for its new Aviation and Air Security Division. As of July 2026, GAO says the agency had fully implemented three, partially addressed four, and taken no steps on one. That earlier report (GAO-25-108568SU) is not public; the status comes from this report's notes.
  • In 2025 the Secret Service created a Protective Readiness section, and its February 2026 guidance calls for reviewing incidents continuously to spot recurring issues. GAO notes that neither that guidance nor the May 2026 incident-reporting policy requires a rationale for keeping the status quo.
  • GAO's Recommendation 1 asks for exactly that line on the standard form. DHS concurred and said the Secret Service will require the "remedial activities" section of incident after-action reports to include the rationale when no policy change is warranted.
  • Eight of 22 advance-planning policies missed their required four-year review. One, for the Vice President's foreign advances, dates to July 2001.
  • The 1991 memorandum that splits protective duties between the Secret Service and the State Department's Diplomatic Security Service has never been updated, though it calls for annual review. It doesn't address drones, and it still describes the Secret Service as part of Treasury. The agency moved to Homeland Security in 2003. Both departments concurred with updating it.

Nathan Tranquilli, GAO's acting director for homeland security and justice, said in the Associated Press's coverage that the drone incidents were a "compelling example" of why the missing documentation matters.

What I could not confirm

What was actually decided in 2015 and 2016. That's the point of the report: there is no record. I have not speculated about it, and neither does GAO.

Whether documentation would have changed Butler. GAO says the information could have shown an emerging trend. It doesn't claim causation, and I don't either.

The full 2025 report. GAO-25-108568SU is a sensitive, non-public product. I relied on this report's summary of its recommendations and their status, and I don't know which one remains unaddressed.

Details of the Butler drone beyond GAO's sentence. Press accounts describe the aircraft and its flight path in more detail. I have kept to what GAO states.

Not my research. I had no part in this audit. My published work is in microwave spectroscopy; the counter-IED comparison above is from my own service, not from the report.

The signal

GAO found the Secret Service met drones near protectees three times between 2015 and 2021, changed nothing about civilian drones, and wrote down no reason, because its incident form never asked for one. Across eleven years it kept the status quo after 58 of 83 incidents with the same blank. The fix GAO asked for is one sentence per incident, and DHS has agreed to it. The lesson travels well beyond the Secret Service: a log that only records the times you acted will never show you the threat you kept deciding to tolerate.

Sources

  1. U.S. Government Accountability Office, Secret Service: Protection Policies Should be Consistently Updated to Better Ensure Protectee Safety, GAO-26-108455, 3 September 2026 (full report online). (PRIMARY. Product page and the full online report read, including the incident findings, the drone sidebar, the Butler paragraph, conclusions, all four recommendations, agency comments, and notes 3, 4 and 42 to 47. Source of the 83/84 and 25 incident counts, the missing-form note, the three drone incidents, the 11-minute Butler flight, the quoted conclusion, the Protective Readiness guidance, the 8-of-22 policy count and 2001 policy, the 1991 memorandum, DHS's planned after-action change, and the July 2026 status of the July 2025 recommendations. The figure of 58 no-change incidents is my subtraction.)
  2. Jesse Bedayn, Associated Press, "Report finds holes and delays in how Secret Service updates its protection policy," via News4JAX, 3 September 2026. (Opened and read. Source of the Tranquilli "compelling example" remark.)
  3. Onur Oncer, The Signal Report: "Why some bases can't stop drones overhead" (244) and "What police need before they can stop a drone" (133). (Earlier reports on counter-drone authority.)

Scope note: this report summarizes a public federal audit at a general level. It contains no information about protective methods beyond what GAO published and no instructions for evading any security measure.

Onur Oncer
Onur Oncer

U.S. Army combat veteran (Counter-IED / Electronic Warfare), peer-reviewed researcher in microwave spectroscopy, and founder & CEO of Shroombiosis. Consults on laboratory operations, AI, and supplement formulation.

← All reports