OpenAI has identified an Iranian operation utilizing its AI tools for disseminating false information about the U.S. presidential election. Responding to its findings, OpenAI has proactively suspended multiple accounts connected to this campaign.
Accounts Banned
Ben Nimmo, a principal investigator at OpenAI, indicated that the accounts related to the Iranian scheme have been terminated. The campaign failed to garner genuine engagement, with Nimmo stating, “We did not see signs that it was getting substantial engagement from real people at all.”
The increasing use of tools like ChatGPT has raised alarms about their potential misuse in spreading falsehoods, especially during elections. In May, OpenAI reported thwarting five other influence campaigns aimed at swaying public opinion and geopolitical dynamics. These campaigns were affiliated with state actors and private entities from countries including Russia, China, Israel, and Iran.
Campaign Tactics
OpenAI labeled the Iranian campaign as Storm-2035. The campaign used ChatGPT to generate content around various U.S. presidential candidates and topics like the Gaza conflict and Scottish independence, often covering both progressive and conservative viewpoints. They produced articles, social media posts, and commentary, but most of the posts received minimal engagement.
Microsoft contributed to the investigation through a Threat Intelligence report released the previous week. The report classified Storm-2035 as part of a wider campaign intended to influence U.S. elections, active since 2020. Microsoft described Storm-2035 as an Iranian network managing multiple sites that impersonate news outlets, engaging U.S. voter groups with polarizing content about presidential candidates, LGBTQ rights, and the Israel-Hamas conflict.
Fake News Outlets
OpenAI pointed out five websites masquerading as credible news sources, with names like “evenpolitics.com”. The group used ChatGPT to write several long-form articles, including one falsely claiming that “X censors Trump’s tweets.”
On social media, OpenAI discovered a dozen X accounts and one Instagram account connected to the operation. The campaign used ChatGPT to rewrite political commentary for these platforms. One tweet falsely alleged that Kamala Harris blames “increased immigration costs” on climate change, using the hashtag “#DumpKamala”.
Last Updated on November 7, 2024 3:15 pm CET
