OpenAI Disrupts AI-Enabled Influence Operations

OpenAI announced on May 29, 2024, that it had disrupted two separate AI-enabled influence operations. These operations utilized sophisticated techniques, including the creation of fake personas and the establishment of a seemingly legitimate think tank, to spread geopolitical messaging. The first operation, which OpenAI has dubbed "Storm-1678," was active from March 2024 and involved the creation of approximately 100 fake personas. These personas were used to generate and disseminate content across various social media platforms, including X (formerly Twitter), Facebook, Instagram, and TikTok. The content was primarily aimed at audiences in the United States and Europe, focusing on themes related to the ongoing conflict in Ukraine and domestic political issues within the United States. The operation also employed AI-generated imagery to enhance the credibility of its fake personas.

The second operation, identified as "Storm-1749," was active from April 2024 and targeted audiences in India and other Hindi-speaking regions. This operation created a network of approximately 30 fake personas, which were used to publish content on X and Facebook. The messaging from Storm-1749 focused on geopolitical narratives concerning India and its neighboring countries, as well as domestic Indian politics. Notably, this operation also involved the creation of a fake think tank called the "Global Strategic Institute," which published articles and reports designed to lend an air of authority to its propaganda. OpenAI stated that both operations were disrupted by taking action against the associated OpenAI accounts and by notifying the platforms where the malicious content was distributed. The company emphasized that these operations were relatively unsophisticated in their use of AI, primarily employing it for text generation and image creation, rather than advanced AI agents or complex autonomous systems. OpenAI's threat intelligence team identified these operations through its ongoing efforts to monitor and combat malicious use of its AI models. The company's policy prohibits the use of its services for deceptive or manipulative purposes, and it actively works to enforce these policies by detecting and disabling accounts that violate them. This disruption highlights the evolving landscape of online influence operations and the role that AI can play in their execution, even at relatively early stages of AI development. OpenAI's proactive stance in identifying and reporting such activities is crucial for maintaining the integrity of online discourse and mitigating the spread of disinformation.