OpenAI has banned a cluster of ChatGPT accounts linked to an Iranian influence operation that was generating content about the U.S. presidential election, according to a blog post Friday. The company says the operation generated AI-generated articles and social media posts, but they do not appear to have reached a large audience.
This isn’t the first time OpenAI has banned accounts linked to state-sponsored actors that use ChatGPT maliciously. In May, the company halted five campaigns that used ChatGPT to manipulate public opinion.
These episodes are reminiscent of state actors using social media platforms like Facebook and Twitter to influence previous election cycles. Now, a similar group (or perhaps the same group) is using generative AI to flood social channels with misinformation. Similar to social media companies, OpenAI appears to be adopting a Whack-a-Mole approach, immediately blocking accounts that appear to be involved in these efforts.
OpenAI said its investigation into this group of accounts was informed by a Microsoft Threat Intelligence report released last week that identified the group (dubbed Storm-2035) as part of a broader campaign to influence the U.S. elections that has been underway since 2020.
Microsoft said Storm-2035 is an Iranian network that imitates news outlets and “actively engages with groups of American voters on opposite sides of the political spectrum with polarized messaging on issues such as the U.S. presidential candidates, LGBTQ rights, and the Israel-Hamas conflict.” As other operations have demonstrated, the playbook is not necessarily about promoting a policy, but rather about fomenting dissent and conflict.
OpenAI identified five website fronts for Storm-2035, representing them as liberal and conservative news outlets with convincing domain names like “evenpolitics.com.” The group used ChatGPT to draft several long-form articles, including one claiming that “X censors Trump’s tweets.” Elon Musk’s platform certainly didn’t do that (Musk is encouraging former President Donald Trump to get more involved with X).
On social media, OpenAI identified 12 X accounts and one Instagram account controlled by the operation. The company says ChatGPT was used to rewrite a variety of political comments, which were then posted on these platforms. One such tweet falsely and confusingly claims that Kamala Harris is attributing the “increased cost of immigration” to climate change, followed by “#DumpKamala.”
OpenAI said it saw no evidence that Storm-2035’s article was widely shared, and most social media posts received few likes, shares, or comments. This is often the case for such tasks, which can be started quickly and cheaply using AI tools like ChatGPT. As the election approaches and partisan bickering intensifies online, expect to see more announcements like this.
