OpenAI cracks down on Iranian influence campaign

Despite low audience engagement, OpenAI has shared threat intelligence with stakeholders and remains committed to detecting and mitigating such abuses through collaboration and advanced AI tools.

OpenAI has intensified its efforts to prevent the misuse of AI, especially in light of the numerous elections scheduled for 2024. The company recently identified and turned off a cluster of ChatGPT accounts linked to an Iranian covert influence operation named Storm-2035. The operation aimed to manipulate public opinion during the US presidential election using AI-generated content on social media and websites but failed to gain significant engagement or reach a broad audience.

The operation generated articles and social media comments on various topics, including US politics, global events, and the conflict in Gaza. The content was published on websites posing as news outlets and shared on platforms like X and Instagram. Despite their efforts, the operation saw minimal interaction, with most posts receiving little to no attention.

OpenAI’s investigation into this operation was bolstered by information from Microsoft, and it revealed that the influence campaign was largely ineffective, scoring low on a scale assessing the impact of covert operations. The company remains vigilant against such threats and has shared its findings with government and industry stakeholders.

OpenAI is committed to collaborating with industry, civil society, and government to counter these influence operations. The company emphasises the importance of transparency and continues to monitor and disrupt any attempts to exploit its AI technologies for manipulative purposes.