A due diligence lesson in PEP-hunting from the BBC

Companies identify politically exposed persons or PEPs mainly through databases compiled from open intelligence sources: published interviews, government directories, news reports, vital records, social media, and so on. But as the BBC showed earlier this month, there’s a powerful new tool for identifying PEPs: facial recognition software.

A team of reporters from the BBC decided to investigate widespread claims that Russian President Vladimir Putin surrounds himself at events with friends, business associates, and even actors, who pose as ordinary local residents.

For example, a blonde woman allegedly appeared with Putin on a fishing trip in 2016 and at an unrelated church service a year later.

Using facial recognition software, the BBC team found images of the blonde woman from the events held in 2016 and 2017 that generated a 99.1 percent match probability.

“Usually, a similarity score of 75 percent or higher should be considered when looking for an identity match,” Professor Hassan Ugail, Director of the Center for Visual Computing at the University of Bradford, told the BBC.

The team then searched Russian media and identified the woman as Larisa Sergukhina, a member of the Novgorod regional parliament from Putin’s United Russia Party. With further research, the BBC team determined the woman heads a fish-trading company and serves as deputy general director of a local agricultural conglomerate.

Several employees working for Larisa Sergukhina appear in photos with her and Putin in various settings. The BBC learned from social media accounts that two of the employees in the photos are father and son.

That’s good reporting.

But is it practical for companies and providers to use facial recognition software to identify potential PEPs?

I did a quick experiment.

I started with a Kremlin-posted group photo of Putin visiting the International Sambo and Boxing Center in Luzhniki, where an older red-haired woman appeared.

I used a free version of facial recognition software called PimEyes to search for other photos of the red-haired woman. In 3.2 seconds, the software returned six results, three of which almost certainly matched the same woman.

The three matches came from attributed Russian media sources. If I wanted to know more about the red-haired woman as a potential PEP, I could continue to the cited sources and beyond.

Fortunately for PEP hunters, political leaders love photo ops. Top politicians in most countries now post near-daily videos and photos of themselves out pressing the flesh. They run dedicated websites, Twitter accounts, YouTube channels, Instagram accounts, and so on. No doubt plenty of supporting characters appear repeatedly in those videos and photos.

So it’s true. There’s another tool in the due diligence toolbox. Facial recognition could unmask droves of yet-undiscovered PEPs hiding in plain sight.