tobias-tullius-4dKy7d3lkKM-unsplash.jpg
Credit: Photo by Tobias Tullius on Unsplash

News organisations must be aware of the narrowing gap between the worlds of Big Tech and politics.

Non-profit newsroom The Bureau of Investigative Journalism, in partnership with Follow The Money and Paper Trail Media, recently published a major, year-long investigation exploring how Russia is using facial recognition technology to identify dissidents.

In an episode of the Journalism.co.uk podcast, Big Tech reporter Niamh McIntyre explained how the story's roots were on Reddit and YouTube where people posted about how to use a little-known platform called Toloka.

@thebureauinvestigates

Russia’s surveillance technology is being unknowingly powered by outsourced workers in the global south. But few are aware what that their work might be contributing to … #russia #tech #putin #surveillance #AI

♬ Suspenseful and tense orchestra(1318015) - SoLaTiDo

Toloka is owned by Yandex, Russia’s biggest tech company, often called "Russian Google". Until recently, it was closely linked with Western financial institutions and had a lot of Western investment. After Russia invaded Ukraine, Yandex has been strong-armed to sell up on unfavourable terms and it is now entirely under Russian ownership. As Vladimir Putin's presidency has progressed, Yandex's ties with the Kremlin have grown stronger.

Toloka recruits gig workers to complete simple tasks for other companies. Gig workers in the global south were being commissioned (for small amounts of money) to complete strange tasks like drawing boxes around humans in CCTV footage or sending pictures of their own faces. The companies processing this data are Tevian and NTechLab, facial recognition tech companies that are in business with the Russian government.

The bottom line is that gig workers in the global south were unknowingly improving Russian surveillance systems used for undemocratic purposes.

"With any AI system, you need large amounts of data," explains McIntyre.

"And that data needs to be organised and labelled in a way that the machine can learn and understand. Often that involves a lot of manual human labour. So for facial recognition, which is used to compare faces that appear on a camera with a dataset of wanted faces, that would often involve tagging images of faces in some way, or uploading images of faces, because you need lots of images for datasets to learn."

This investigation was possible because the information was largely open-source, albeit with the use of Russian VPNs. But that could change as the internet becomes more divided. This has been dubbed the "Balkanisation" of the internet, or the "splinternet".

"So you're going to have this Russian internet, a Chinese internet and a Western internet. Companies like Yandex, [given] the fact that it's had to pull back from the international aspects of its business, certainly do fit into that theme."

McIntyre, also an AI accountability fellow at the Pulitzer Centre, says that authoritarian regimes leveraging technology - "digital authoritarianism" as it is known - is a theme that will continue. But her attention may well turn to Western democracies in the future - and that will make it harder for journalists to gain access to sources.

"Facial recognition is used pretty widely in the US, not as undemocratically as it is in Russia, but there's been a lot of questioning the legitimacy of certain uses of it.

"In the US, the links between US AI companies and the US defence industry are becoming increasingly prevalent. As the defence sector instrumentalises AI, they are turning to tech companies to help them fulfil those needs."

Join the conversation with us at the digital journalism conference Newsrewired on 22 May 2024. Check out the full agenda and grab your ticket now

Free daily newsletter

If you like our news and feature articles, you can sign up to receive our free daily (Mon-Fri) email newsletter (mobile friendly).