OpenAI has banned accounts originating from Russia that used ChatGPT to promote a fake Israeli academic institute. The network copied research and attributed it to academics who did not write it, aiming to spread pro-Russia narratives.
The sophisticated use of AI by Russian-linked networks to create and disseminate disinformation, even when targeting smaller audiences, highlights the evolving tactics of hybrid warfare and the challenges in detecting and countering such operations.
A network originating in Russia utilized ChatGPT to promote an organization masquerading as an academic institute, according to a report by OpenAI. The company stated it banned a group of accounts that employed Russian-language prompts to generate social media content and comments, primarily in English, while attempting to conceal their Russian origin.
OpenAI noted that while the campaign's reach was limited, its sophisticated structure set it apart from other Russian-linked influence operations. Due to ChatGPT's unavailability in Russia, the operators reportedly used VPNs to access the service. The network promoted the International Burke Institute, which claimed to be an Israel-based expert group, through official-looking profiles and user accounts. OpenAI's review found that 34 out of 36 articles on the institute's website were copied, with some incorrectly attributed to academics.
This incident is part of a broader trend of generative AI being used for political influence campaigns. OpenAI previously identified a separate Russia-linked campaign, Doppelganger, which used ChatGPT to create content favorable to Russia and critical of Ukraine, NATO, and the United States. Additionally, groups linked to Russian military intelligence, such as Forest Blizzard, have used ChatGPT for cyber operations, including target research and phishing attacks.
While the operators used Russian prompts, OpenAI did not directly attribute this specific campaign to the Russian government. The report also indicated that some individuals in Israel may have represented the institute at conferences or through submissions, but their connection to the Russian-based operators remains unclear.