More than two dozen members of Congress asked the Federal Aviation Administration Tuesday to stop using a new artificial intelligence tool that helps manage air traffic around Washington, D.C., saying they have seen no safety evaluation or testing reports for it, Reuters reported. The request is one of the first direct challenges to the use of AI decision-support software in a safety-critical U.S. system, and it lands as the FAA plans to roll the tool out nationwide.

What Happened

The lawmakers' letter, signed by Democrats including Reps. April Delaney, Alexandria Ocasio-Cortez, Don Beyer, Jamie Raskin, Raja Krishnamoorthi and Kweisi Mfume and Sen. Chris Van Hollen, raises concerns about what it calls a pilot project. According to Reuters, the signers say they have not seen safety evaluations or testing reports, and have not been told of any effort to train aviation safety staff on the system. They asked the FAA to respond by Oct. 19. Reuters did not report a response from the FAA.

The tool is called Strategic Management of Airspace, Routes, and Trajectories, or SMART. As Reuters described it, the software uses predictive analytics, statistical models that forecast what is likely to happen, to draw on airline schedules, weather, airport capacity and airspace conditions. It predicts traffic flows and flags potential conflicts before they arise. The Transportation Department says it uses more than 200 data streams, produces visualizations and offers scheduling recommendations for FAA operations experts to consider, and that it does not actively manage aircraft in flight.

RelatedWhite House AI Task Force Plans Breach Warnings From Labs at Its First Meeting

The FAA awarded the work to Air Space Intelligence in June under a 12-year contract worth $875 million and began using the tool last month in airspace around Washington. Its first job is rescheduling flights canceled because of bad weather. The FAA plans to expand it gradually to other parts of the country and to tasks such as coordinating flight schedules and routes before aircraft take off.

Transportation Secretary Sean Duffy defended the system on Fox Business, calling it essentially "dumb AI" and saying controllers will always control the airspace. FAA Administrator Bryan Bedford told Reuters the tool would be "transformational" and would help the agency make better decisions.

Why It Matters

The dispute turns on a question many organizations face as they adopt AI: what evidence should be required before software that makes recommendations is trusted inside an operation where mistakes are costly. Officials stress that SMART only advises and that people make the final call. The lawmakers' objection is that no safety testing has been shown to them and that the staff responsible for safety have not been trained on it. Even an advisory tool shapes decisions, and its users need to know when to distrust it.

Washington's airspace carries particular weight. On Jan. 29, 2025, an Army helicopter and a regional passenger jet collided near Ronald Reagan Washington National Airport, killing all 67 people aboard. The National Transportation Safety Board's investigation faulted the FAA's helicopter route design and its failure to act on earlier warnings, along with high controller workload. Reuters' account of the letter does not mention the crash.

For companies running critical infrastructure, from utilities to logistics networks, the episode signals that elected officials may ask for documented testing and staff training before AI tools are allowed to shape operational decisions, even when a human keeps final authority. Vendors selling AI decision aids to government should expect similar demands for safety evidence.

What to Watch

The FAA's reply, due Oct. 19 by the lawmakers' timeline, will show whether it has safety evaluations to share or intends to keep expanding the tool. Watch also for any move to bring SMART into new regions or into pre-departure route planning, and for whether Republican lawmakers or aviation unions join the call.