Princeton researchers develop PACMAN AI to stabilize fusion systems quickly

Scientists at the Princeton Plasma Physics Laboratory and Princeton University have developed an AI framework called PACMAN that can make critical safety decisions for fusion energy systems in milliseconds.

The breakthrough addresses a major challenge in fusion research: controlling plasma particles that can reach temperatures hotter than the sun’s core. These particles can destabilize in mere thousandths of a second—far faster than human operators can respond. Traditional control systems struggle to keep up, risking damage to fusion reactors.

PACMAN uses artificial intelligence to monitor and adjust fusion systems in real time, ensuring stability while maintaining human oversight of broader operational goals. The framework acts autonomously only in split-second scenarios, intervening to prevent disruptions that could harm the machine. Researchers say this balance of AI speed and human control could accelerate the development of safe, practical fusion energy.

The innovation comes as global efforts to harness fusion as a clean energy source intensify. With PACMAN, scientists move closer to overcoming one of the biggest technical hurdles—keeping superheated plasma stable long enough for sustained energy production. Next steps include testing the system in more complex fusion environments.