WASHINGTON, October 15, 2026 – Air traffic controllers already have to operate in one of the most demanding environments conceivable: busy skies, variable weather conditions, and quick decisions. Now, a new artificial intelligence program deployed to manage flight traffic near Washington’s busiest hubs has triggered a sharp backlash from more than two dozen congressional Democrats.
The lawmakers sent an urgent letter demanding that the Federal Aviation Administration cease using the AI system that was introduced last month at Reagan National, Dulles International, and Baltimore-Washington International Airports, citing a total lack of safety evaluation reports or formal training for controllers.
“We are deeply concerned about the integration of any AI-enabled system into air traffic control systems, where decisions are literally a matter of life or death,” the lawmakers wrote to Transportation Secretary Sean P. Duffy and FAA Administrator Bryan Bedford.
What the Tool Does
To be clear, this isn’t an independent controller.
The software crunches airline schedules, weather data, and runway congestion to predict traffic bottlenecks from the FAA’s command center in Warrenton, Virginia. It then feeds real-time recommendations directly to air traffic controllers. “Decision support might sound harmless on paper, but algorithmic suggestions still heavily sway human choices when stress levels spike,” noted one regional aviation safety consultant.
The concern is straightforward: what happens when a recommendation is wrong, incomplete, or based on faulty inputs?
Why Lawmakers Are Pushing Back
In their letter, Rep. April McClain Delaney from Maryland, Rep. Don Beyer from Virginia, Rep. Alexandria Ocasio-Cortez from New York, and Sen. Chris Van Hollen from Maryland emphasized that regulators haven’t produced a single public safety assessment for the software. Controllers, they added, haven’t received proper training on how to handle algorithmic glitches.
Another letter written by 23 Democrats addressed to the Government Accountability Office warned that predictive software could distort an experienced controller’s judgment or even “hallucinate safety-critical information.”
The latter issue is no hypothetical one. Generative and predictive AI systems can come up with inaccurate but confident results, and the culture of aviation is based on redundancy and verification. So the problem lies not in AI’s ability to assist; the question is how the FAA will be able to prove that AI assists without adding any new modes of failure to its work.
FAA’s Position—and Its Silence
FAA declined to comment on the issue.
This silence creates a notable vacuum – there’s no official reaction to the allegations of missing evaluations or training raised by the lawmakers.
According to Transportation Secretary Duffy, AI technology would never take over human controllers in the management of the national airspace. However, the department intends to deploy the tool beyond the Washington area, which is the reason why its critics argue for a pause in its application.
Here’s the tension.
AI could be used by controllers to predict air traffic problems, bad weather, and scheduling conflicts in advance. But without transparent testing, clear security measures, and training of controllers, such a potentially useful tool could become just one more element of the opaque system.
What Happens Next
The immediate question is whether the FAA will hit pause, release its safety documents, or push forward regardless. The GAO inquiry adds serious federal heat.
For travelers flying through D.C., the bottom line is simple: they want shorter delays without compromising safety. For the FAA, proving that an algorithm is as trustworthy as the human in the tower just became a whole lot harder.
