AI Agents Could Soon Be Fighting The Wars Of The Future

Artificial intelligence agents are taking part in simulated cyber warfare alongside human military operators as a NATO-accredited defense center prepares for a future in which machines could help wage war.

The Crossed Swords 2026 exercise, which began Monday in Tallinn, Estonia, brings together more than 250 participants from over 40 countries to coordinate cyber operations with conventional military action, according to the NATO Cooperative Cyber Defence Centre of Excellence (CCDCOE). For the first time, participants are developing and deploying AI agents to operate alongside human military specialists, the center announced.

The five-day exercise is running from Monday through Friday and includes military cyber operators, digital forensics analysts, threat hunters and command personnel, according to the NATO CCDCOE.

Unlike conventional cybersecurity training, Crossed Swords requires military personnel to coordinate digital attacks with physical battlefield operations, including special operations missions, the center explained.

The objective is to synchronize cyber operations with conventional military action, potentially allowing forces to exploit enemy network vulnerabilities at decisive moments during combat, according to the CCDCOE.

Timing is particularly important because cyber operations must produce their intended effects within minutes of physical military action rather than hours, exercise director Jozef Kostelansky said in the official announcement.

“AI is now in the loop on both sides,” Kostelansky said, according to the CCDCOE.

The exercise represents a potential shift toward warfare in which artificial intelligence performs portions of military cyber operations while human personnel oversee the broader mission. However, the organizers have not established that these systems can independently authorize military attacks or select targets without human approval.

The exercise is intended to integrate AI into military operations without sacrificing “critical human oversight,” Kostelansky said, according to the CCDCOE. The technological environment includes locally hosted artificial intelligence systems and, for the first time, OpenAI’s Daybreak cybersecurity technology.