September 1983: Soviet officer defies military protocol, averting potential nuclear war

About a quarter past midnight on Sept. 26, 1983, a computer panel lit up at a top-secret control center south of Moscow, where the Soviet military monitored signals from its missile defense early warning satellites. The panel displayed the word “LAUNCH” in red letters while a siren blared. The alert indicated that the United States had just launched an intercontinental ballistic missile from an air force base in Montana. If it were true, the missile would reach Soviet territory in about 30 minutes.

Lieutenant colonel Stanislav Petrov was filling in as the officer in command that night, evaluating satellite signals for potential launches to report up the leadership chain. He was shocked; an alert like this had never come before.

A few minutes later, the system sounded alerts for four more missiles. As it went into overdrive and his staff scrambled, Petrov ran through possibilities in his mind. But time was running out. Still not entirely sure, he made a decision: It was a false alarm. He reported to headquarters that the computer system had malfunctioned.

After about 25 agonizing minutes, his report was proven right — but months would pass before Soviet investigators understood the strange cause of the erroneous alert.

At that time, tensions between the U.S. and USSR were at a peak. Earlier that year, President Ronald Reagan had deemed the Soviet Union an “evil empire” and announced the Strategic Defense Initiative, nicknamed the “Star Wars” program, to intercept incoming nuclear missiles. Three weeks before the alert, the Soviet Air Forces had shot down Korean Air Lines Flight 007 when it strayed into Soviet airspace, killing everyone on board — including 62 Americans, among them a U.S. member of Congress. With both nations on high alert, it was plausible that the U.S. would launch a nuclear attack on Moscow.

The Soviet military used a space-based early warning system named Oko, or “eye” in Russian — a constellation of satellites trained to identify ballistic missile launches by sensing the infrared radiation emitted from their rocket plumes. These satellites operated in Molniya orbits — highly elliptical trajectories that swung from altitudes as low as 500 kilometers to as high as nearly 40,000 kilometers above Earth, spending roughly eight hours of their 12-hour orbits near the highest point, or apogee. From there, they monitored high-altitude regions on Earth — particularly, missile fields across the Northern Great Plains, where the Soviets knew ballistic missiles would launch from underground silos if the U.S. carried out an attack.

The infrared sensors on Molniya satellites pointed downward at an angle, “grazing” the Earth’s horizon and using the black backdrop of space to minimize thermal noise while scanning for exhaust plumes breaking through clouds in the atmosphere. Compared with the top-down view of U.S. early warning systems’ geosynchronous satellites, the angle required by Molniya satellites was much more complex, requiring intricate adjustments and leaving sensors vulnerable to optical illusions.

But it would have been difficult to predict the rare geometric alignment that caused the Oko system to malfunction on Sept. 26. Just days after the autumnal equinox, the sun sat low on the horizon relative to where the satellite Kosmos-1382 monitored the late afternoon skies above the Northern Plains. Clouds composed of tiny ice crystals sat high above the plains, acting as a “mirror” that reflected the sun’s rays at an incidence angle aligned with the satellite’s grazing angle — in such a way that beams of sunlight were sent straight into the satellite’s infrared sensors. The computer system interpreted this specular reflection as the thermal signature of a missile plume, setting off alerts.

Soviet leadership assembled a team to investigate what had happened. Petrov was quickly reprimanded for failing to fill out paperwork and reassigned to another post. He took early retirement in 1984, after which he lived on a small military pension and, at one point, had to grow potatoes to make ends meet. Because the incident was classified until after the fall of the Soviet Union, he would not share his account of that night until more than 15 years later.

By around 1999, he began conducting interviews about his experience with international media, gaining fame and becoming known as “the man who saved the world” — what became the title of a 2013 docudrama that recounts his story.

Petrov, a civilian engineer by training, was the deputy chief of the Department for Combat Algorithms. Most days, he worked behind the scenes refining Oko’s software and responding to system glitches. Having helped build the system, he knew that it had been rushed into active service only a year earlier, its underlying flaws left unaddressed. Without confirmation from the ground-based radar, he could not bring himself to trust the alert. He also reasoned that a real U.S. strike would never arrive as five isolated launches, but as an overwhelming barrage of hundreds of missiles. With that logic in mind, he trusted his gut over the computer — a decision that ultimately cost him his career.

“That was my job,” Petrov said in a 2013 BBC interview, responding to claims that he had done something particularly heroic. “But they were lucky it was me on shift that night.”

In the aftermath, the Soviet military moved to address Oko’s vulnerabilities. They adjusted protocols, worked to improve the software, and accelerated the deployment of geostationary satellites — which hovered at a fixed spot above Earth, providing continuous coverage of the land below — to supplement the Molniya satellites.

But for the remaining years of the Soviet Union and beyond, the incident would remain buried — and with it, another revelation: When high-stakes decisions are entrusted to automated machines, human judgment may remain the ultimate safeguard.