Reports that more than 400 unmanned aerial systems were observed at or near U.S. military installations have grabbed attention because the figure crystallizes two anxieties at once: that our bases are increasingly exposed, and that the government does not yet have coherent tools to manage the problem. The public record, however, is more nuanced. The senior NORAD and U.S. Northern Command commander told lawmakers that there were 350 detections of small UAS at roughly 100 installations in 2024. That baseline is the best publicly attributable figure for last year, and it helps explain why claims of “hundreds” of incursions feel credible even if precise tallies vary by source.

Numbers matter, but so do definitions. A “detection” can cover everything from a confirmed small commercial quadcopter to a radar return later determined to be a flock of birds or a misreported manned aircraft. In December 2024 federal agencies publicly said many of the thousands of civilian drone reports reviewed around that period were a mix of lawful drones, misidentified aircraft and even astronomical objects. That does not negate the security problem, but it does complicate headline counts and the assignment of operational intent.

Operational impact is the crucial lens. Some incursions have had measurable consequences. The Langley episode in December 2023, when repeated nighttime sightings around a major fighter base forced operational adjustments, became a touchstone for Pentagon leaders arguing that detection gaps translate into real-world vulnerability. At congressional hearings and in Pentagon briefings, officials have emphasized not only frequency but also the capacity of some UAS operations to surveil sensitive assets and exploit gaps in detection and attribution. Those assessments have driven an urgency to accelerate counter-small UAS efforts.

Why the military struggles with these incursions is partly technological and partly legal. Many U.S. installations lack the layered sensor suites that would reliably detect low-signature, low-altitude UAS operations across long timelines. Even when sensors pick up activity, rules of engagement and interagency authorities can slow response. Military leaders have pointed to both the need for better hardware and for clarified authorities so base commanders can act quickly when they face a suspicious small UAS over a perimeter. Exercises like Falcon Peak are explicitly designed to field commercial and military vendors against realistic low-signature and distributed UAS profiles to close those gaps.

Attribution remains the hardest problem. The available public reporting shows a mix of actors and causes: hobbyists flying near restricted airspace, commercial operations that stray off approved routes, law enforcement or media flights, criminal or cartel smuggling activity in border regions, and genuine concerns about foreign intelligence collection. In some cases abroad, U.S. and allied bases have seen flights that security services judged to be deliberate surveillance. In others, post-incident review has found no hostile intent. That heterogeneity means policy responses have to be layered and discriminating, not one-size-fits-all.

What should the Department of Defense and partner agencies prioritize now? First, a mapped inventory of which installations have end-to-end counter-UAS capability, and where gaps create operational risk. Second, rapid fielding of interoperable sensor suites and defeat options for the most critical sites, coupled with carefully circumscribed legal authorities that let commanders act when there is reasonable suspicion of hostile surveillance or attack. Third, better civil-military coordination with the FAA, Department of Justice, state and local law enforcement, and industry to separate nuisance incidents from strategic threats quickly. Congress has already signaled interest in documentation and reform, and exercises and pilot fieldings must be matched by funding and policy changes.

There are policy trade-offs that demand attention. Expanding kinetic authorities or jamming capabilities raises risks to commercial aviation, civilian privacy and civil liberties. Overclassifying incident data to avoid alarm will undercut public trust and make it harder to enlist local communities in reporting or mitigation. The correct path is a calibrated three-track approach: harden and defend the most sensitive assets; modernize detection and attribution across the force posture; and build transparent, accountable legal frameworks for response. This is as much a governance problem as a technology one.

Finally, about the headline figure itself. Whether the count reads 350, 400, or some higher aggregated number depends on the window and criteria used. What matters more than the exact integer is the trend and the capability gap it signals. Even conservative, attributable numbers from military testimony show the scale is large enough to require immediate, coordinated action. For policymakers and industry that work on both the hardware and the rulebook, the takeaway is straightforward: treat the problem as enduring, prioritize attribution and interoperability, and avoid binary thinking that assumes every sighting is an imminent attack or else wholly innocuous. The skies over bases will not quiet themselves, so the institutional response must be faster and smarter than the platforms that threaten to exploit current seams.