OpenAI Bans Russian Account Fake Think Tank Network

OpenAI has banned a group of ChatGPT accounts linked to Russia that were running a covert influence campaign. The operation centered on the International Burke Institute, a website that pretended to be an independent Israel-based expert community. OpenAI found that 34 of the 36 articles on the site were copied from other sources, sometimes with wrong authors attached. The site even claimed connections to well-known scholars like Francis Fukuyama and Noam Chomsky to look more credible.

The institute's main product was a "sovereignty index" that rated countries on several measures, including politics, technology, and military power. According to OpenAI, the index was designed to portray Russia in a positive light while criticizing Western governments. The same operators also used ChatGPT to write German-language posts for a Telegram channel that opposed Ukraine and the EU while praising Russia. They even made logos for about a dozen pro-Russian Telegram channels in different countries.

The campaign's reach was small - most posts got few views, and the Telegram channels had only 10,000 to 20,000 followers. But OpenAI says the real danger lies in what the operation was trying to build: a fake institution that looks real enough to make its messages seem trustworthy. Generating a single fake post is easy, but creating an entire fake research organization around it is much harder - and that's exactly what AI now makes possible.

The good news is that AI also leaves traces that investigators can follow. OpenAI said ChatGPT activity helped them connect the dots across the whole campaign. This shows that while AI can make influence operations more sophisticated, it can also make them easier to detect - as long as companies keep looking.