OpenAI Disrupts Iranian Influence Operation Storm-2035
OpenAI has identified and banned a cluster of ChatGPT accounts associated with Storm-2035, a covert Iranian influence operation. This action was taken to prevent the manipulation of public opinion and the influence of political outcomes through AI-generated content during the 2024 election cycle.
Operation Storm-2035 Capabilities and Tactics
Storm-2035 utilized ChatGPT for two primary workstreams to create content across multiple platforms:
- Long-form Content: The operation produced articles on U.S. politics and global events, which were published on five websites posing as both progressive and conservative news outlets.
- Short-form Content: The operation generated short social media comments in English and Spanish. This included asking AI models to rewrite existing comments posted by other users to blend in.
To maintain a facade of authenticity, the actors interspersed political messaging with non-political content regarding fashion and beauty.
Target Topics and Geographic Focus
Storm-2035 focused its content generation on several key geopolitical and social issues, targeting both sides of the political spectrum:
- U.S. Presidential Election: Commentary on candidates from both progressive and conservative perspectives.
- Global Conflicts: The conflict in Gaza and Israel’s presence at the Olympic Games.
- Regional Politics: Politics in Venezuela, Scottish independence, and the rights of Latinx communities in the U.S. (produced in both English and Spanish).
Impact Assessment and Engagement Metrics
Despite the scale of the operation, the activity did not achieve meaningful audience engagement. Using the Brookings Breakout Scale’s impact assessment (where 1 is lowest and 6 is highest), OpenAI categorized this operation at the low end of Category 2. This indicates that while activity occurred across multiple platforms, there was no evidence that real people widely shared or picked up the content.
Most identified social media posts received few or no likes, shares, or comments, and web articles were not widely shared across social media platforms.
Mitigation and Industry Collaboration
OpenAI has banned the involved accounts and continues to monitor for further policy violations. To support the wider community in disrupting similar activities, OpenAI shared threat intelligence with government, campaign, and industry stakeholders. This investigation was aided by information previously published by Microsoft.
OpenAI states that it is leveraging its own AI models to better detect and understand abuse and remains committed to partnering with industry, civil society, and government to uncover and mitigate covert influence operations at scale.