OpenAI has pulled the plug on a number of ChatGPT accounts originating from Russia, accusing them of running a sneaky online misinformation campaign.
The company, led by Sam Altman, stumbled onto the scheme while digging into AI-generated social media posts, which led them to uncover a larger influence operation, according to an OpenAI blog post on Tuesday.
The whole thing revolved around a website that hosted stolen and misattributed academic papers, a 'sovereignty' index that gave a thumbs-up to Vladimir Putin's Russia while trashing Western countries, and a bunch of attempts to hide the fact that the operators were Russian. The folks behind it used Russian prompts to churn out social media comments on platforms like Substack, Telegram, Meta Platform Inc.'s (NASDAQ:META) Facebook, and LinkedIn. They even told ChatGPT to scrub any linguistic hints that they were from Russia.
Most of the content pushed an outfit called the International Burke Institute (IBI). The IBI's website is a patchwork of articles lifted from legitimate academic sources, sometimes with the wrong authors attached. The operation also cooked up a 'sovereignty index' that sang Russia's praises and took shots at the West.
While the operation's immediate impact seems pretty small, OpenAI says the real deal is the infrastructure it built. The AI-generated posts pointed to a seemingly legit institution, complete with fake experts, republished academic work, and a supposed proprietary risk index.
