Feds bust Russia-backed, AI-powered bot farm — more coming

“With these actions, the Justice Department has disrupted a Russian-government backed, AI-enabled propaganda campaign to use a bot farm to spread disinformation in the United States and abroad.” — Merrick B. Garland, United States Attorney General

“Russia intended to use this bot farm to disseminate AI-generated foreign disinformation, scaling their work with the assistance of AI to undermine our partners in Ukraine and influence geopolitical narratives favorable to the Russian government.” — FBI Director Christopher Wray

That was Garland and Wray in July 2024 after the US Department of Justice (DOJ) announced the seizure of two domain names and 968 social media accounts associated with a large-scale disinformation campaign using AI-powered software called Meliorator.

What Meliorator does

Allegedly created by a deputy editor-in-chief at Russian state-sponsored media company RT (formerly Russia Today), Meliorater has been used by RT affiliates for the past two years to create large numbers of social-media bots that use generative AI to escape detection and disseminate disinformation. They have targeted the US as well as Poland, Germany, the Netherlands, Spain, Ukraine, and Israel, according to the DOJ

Each bot has a persona that is defined by a detailed set of parameters—its archetype—that guide its behavior. They’re able to write original posts, like, follow, comment, re-post, and gain followers—all in the service of amplifying malicious Kremlin-directed content intended to sow discord. 

Their choices of whom to follow and what to repost are driven by the stated interests and opinions of their personas. They re-post and amplify messages from their fellow bots, and mostly follow real accounts with at least 100,000 followers, which helps them evade detection. They are also able to obfuscate IP addresses and bypass dual-factor authentication.

They behave, and communicate, like real, fully-fleshed-out people. They share lots of legitimate content, but they also post, re-post, and comment inflammatory content that, according to the DOJ, comes from the Russian Federal Security Service (FSB). And even then, the bots include specific context and framing that reflects their persona and archetype. 

The DOJ, along with other international agencies, issued this Joint Cybersecurity Advisory providing comprehensive technical details of Meliorator and how it works. 

How social media platforms can respond

According to intelligence agencies involved in the operation, social media platforms can reduce their exposure to Meliorator-generated accounts by taking a few steps:

• Implement processes that help ensure that each account is generated by an actual human whose actions conform to the platform’s Terms of Use 

• Audit, and where possible improve, authentication and verification processes

• Create and follow protocols to identify users with known suspicious user agent strings

• Improve account security by imposing MFA, settings to protect privacy, and more by default on all user accounts