OpenAI said it has banned Russian and Iranian accounts that used its artificial intelligence tools to support influence operations targeting multiple countries.
The company disclosed the action in a blog post, saying it investigated the campaigns and found that operations focused on placing false narratives in established media outlets appeared to have reached a wider audience than those relying on fake social media accounts.
The Russian operations targeted countries in Latin America and appeared aimed at undermining support for Ukraine and influencing local politics. The accounts were primarily using ChatGPT to write internal reports detailing their activities to unspecified superiors.
One of those reports described what OpenAI called the widest-reaching influence operation it had encountered so far. In that campaign, Russian operatives said they impersonated a regional educational authority in Peru and tricked schools into organizing activities featuring Stepan Bandera, a controversial Ukrainian nationalist figure admired by some for leading a drive for Ukrainian independence before and during World War II, but reviled by others, including in Poland, because his movement sided with the Nazis and killed Poles and Jews.
A Peruvian outlet reported that one such event did take place at a school. A Polish news report about the event then prompted a far-right Polish member of the European Parliament to suggest banning entry to people who professed support for Bandera.
"The Russian fake thus both fed on, and fed into, historical tensions between Ukrainians and Poles," OpenAI wrote in its report.
The Russian internal memos also claimed that operatives ran a fake think tank in Latin America and recruited unwitting researchers to interview experts and compose reports. OpenAI said that effort was similar to a 2020 Russian operation that posed as an independent news outlet linked to the Internet Research Agency, which was involved in efforts to influence the 2016 U.S. election.
In other instances, the Russian operatives took credit for events they were not responsible for. OpenAI noted that AI was used to generate fake material, but said it represented a minority of the overall workload.
"Russians are constantly engaging in these sorts of active measures," said Darren Linvill, co-director of a media forensics research hub at Clemson University who studies online influence operations. "Trying to engage in narrative laundering, creating fake stories, fake personas to either create or spread propaganda that aligns with Russia's worldview and Russia's political agenda."
Linvill added: "These Russian underlings were trying to use AI to create a shortcut and do less work in the same way that a lot of us use AI, but in this case it helped get them caught."
The Iranian operation was separate in both method and focus. Iranian operatives used AI to help produce content in Persian and English, primarily centered on U.S.-Iran relations and, to a lesser degree, American politics.
OpenAI said the Iranian operation's activity "appeared consistent with a commercial actor running a for-hire influence campaign," but said it was unable to identify who was behind it.
According to OpenAI, Iranian accounts created seven fake author profiles, which were used to pitch and publish lengthy articles in small media outlets around the world. Nearly 100 articles were published in this way. The fake authors placed content in outlets including a progressive U.S. news site and a U.K.-based Middle East publication. The latter has since removed the relevant author pages.
