Intelligence isn’t just for spies. Commercial satellites, AI, and open-source intel are allowing citizen detectives to investigate nuclear secrets.

“The world of open-source nuclear sleuthing is wide open to anyone with an Internet connection,” Amy Zegart writes in Spies, Lies, and Algorithms. “It draws people with a grab bag of backgrounds, capabilities, motives and incentives - from hobbyists to physicists, truth seekers to conspiracy peddlers, profiteers, volunteers and everyone in between.”

There’s no standardized quality control processes, ethical guidelines, or formal training programs, however, and even when the sleuthing is conducted by professionals problems can still arise.

Liam O'Murchu and Eric Chien, two cyber detectives working for California’s Symantec, usually help clients ward off computer viruses. In 2010, they were the first people to investigate Stuxnet, a malicious computer worm believed to have damaged Iran's nuclear program. Even before they finished their investigation, Liam and Eric found themselves in a murky world of international spies, coded threats, and mysterious suicides.

‍

Citizen spies at work ‍

Superpowers and professional cyber spies aren’t the only ones snooping on nuclear secrets. Citizen spies can synthesize information volunteered by users of tracking apps and community data sharing sites while machine learning techniques can analyze massive quantities of data or images. 

“For those analyzing nuclear threats, machine learning can help detect changes over time at known missile sites or suspect facilities,” Zegart said, citing three instances where citizen spies - albeit some of them with considerable expertise - investigated nuclear secrets using open-source intelligence and tools.

‍

Nuclear secrets

In July 2020, two NGO researchers investigated a fire that the Iranian Atomic Energy Organization described as an incident involving an ‘industrial shed’ under construction. Using commercial satellite imagery, geolocation tools, and other data, the NGOs determined that the shed was a nuclear centrifuge assembly building at Natanz, Iran’s main enrichment facility. The Associated Press published a story about their research as did The New York Times.

In a second incident, University of Missouri researchers, at the request of the US National Geospatial-Intelligence Agency, developed machine-learning tools to measure how accurately and quickly they could identify surface-to-air missile sites in southwest China. Using commercially available satellite imagery, they correctly identified 90 percent of the missile sites. The machines took 42 minutes to scan an area of about 90,000 sq km, roughly the size of Maine. 

And finally, in 2012, Stanford nuclear scientist Siegfried Hecker and geospatial analyst Frank Pabian determined the locations of North Korea’s first two nuclear tests using open-source technology to recalculate the epicenters of two nuclear tests. The Stanford researchers used results from seismic signals and Google Earth 3D maps of the test area.

‍

‍

‘An intelligence agency for the people’

Bellingcat’s founder, Eliot Higgins, describes the organization in his book as “an intelligence agency for the people”. On any given day, Bellingcat may publish an article investigating a slowdown at Brazil’s nuclear sub shipyard or US soldiers revealing nuclear weapons secrets via flashcard apps.

Sometimes, even researchers with the best of intentions can find themselves in over their head, however. Journalistic reports from 2001 speculating that Iraqi leader Saddam Hussain may have already tested nuclear weapons using enriched uranium have never been confirmed and the evidence has never been found.

“Using cheap satellite imagery, deep fakes and weaponized social media, foreign governments - or their proxies, or just individuals looking to make trouble - will be able to inject convincing false information and narratives into the public debate,” Zegart writes. 

For the CIA and the other intelligence agencies, this is a seminal moment.

“To succeed, spy agencies will need to operate differently: giving open-source intelligence much greater focus and attention, harnessing new technologies and tradecraft to improve their own collection and analysis, and understanding that open-source intelligence isn’t just intelligence,” Zegart said. 

“It’s an entirely new ecosystem of players with their own motives, capabilities, dynamics and - importantly - weaknesses.”