OpenAI Uncovers Iranian Influence Campaign Using ChatGPT to Sway U.S. Election and Global Politics

OpenAI Uncovers Iranian Influence Campaign Using ChatGPT to Sway U.S. Election and Global Politics
Image credit: Dreanstime

OpenAI recently revealed that it detected an "Iranian influence operation" leveraging ChatGPT to generate content aimed at shaping public opinion in the United States and beyond. This group, identified as Storm-1679, was found to be producing articles and social media posts targeting the 2024 U.S. presidential election, focusing on figures like Vice President Kamala Harris and former President Donald Trump.

Influence Beyond U.S. Politics: Global Reach of the Operation

In addition to meddling in the U.S. election, Storm-1679's activities extended to a variety of global issues. OpenAI reported that the group also generated content related to Israel's invasion of Gaza, the 2024 Olympics, the rights of U.S.-based Latinx communities, Venezuelan politics, and the movement for Scottish independence from the U.K. These efforts highlight the broad scope of the group's influence operations.

Limited Impact Despite Widespread Efforts

While Storm-1679's content creation was extensive, OpenAI noted that most of the posts and articles had minimal engagement from real users. The company detailed the incident on its blog, revealing that it identified and removed a dozen X (formerly Twitter) accounts that posed as both conservative and progressive voices. These accounts used hashtags such as "#DumpTrump" and "#DumpKamala" to disseminate their messages. Additionally, at least one Instagram account was involved in spreading AI-generated content.

Sample of the X post with #DumpTrump and #DumpKamala
Image credit: Gizmodo

OpenAI's Response to the Discovery

Upon discovering the operation, OpenAI took swift action by banning a cluster of accounts linked to Storm-1679. The company also shared its findings with government, campaign, and industry stakeholders, though it did not specify which entities were involved. Screenshots of the posts shared by OpenAI showed minimal engagement, with view counts ranging from just 8 to 207 and few likes or shares.

The Broader Context: Similar Campaigns and Cyber Threats

This incident is part of a larger pattern of online foreign interference in U.S. elections. Microsoft, for example, released a report on August 6 describing another Iranian network, Storm-2035, which created polarizing posts about the election, LGBTQIA+ rights, and Israel’s actions in Gaza. The report also detailed a phishing attack linked to Iran that targeted a high-ranking U.S. campaign official. This type of interference echoes past events, such as the 2016 phishing attack by the Russia-linked group Guccifer 2.0, which leaked thousands of Democratic National Committee emails.

Conclusion: The Ongoing Battle Against Foreign Influence

Under pressure from lawmakers, tech companies have been ramping up efforts to combat foreign influence in elections. These initiatives include fact-checking memes, implementing political ad bans, and forming collaborations with rivals and law enforcement. The discovery of Storm-1679's activities underscores the persistent threat of foreign interference and the challenges faced by tech companies in protecting the integrity of democratic processes worldwide.