OpenAI has taken decisive action against an Iranian group that used its ChatGPT chatbot to generate and disseminate content aimed at influencing the upcoming US presidential election.
The group has been identified as Storm-2035, which produced commentary on key political topics, including the candidates in the tight race between Democratic candidate Kamala Harris and Republican rival Donald Trump. Additionally, this group focused on polarising issues such as the Gaza-Israel conflict and Israel’s participation in the Olympic Games.
Despite the group’s efforts to sway public opinion, the operation failed to achieve traction. OpenAI’s investigation revealed that the majority of the generated content, including long-form articles and shorter social media posts, received minimal engagement, with most posts attracting few or no likes, shares or comments. Further, there were no signs that the articles had been widely shared across social media platforms.
In response, OpenAI has banned the accounts involved in the operation and is closely monitoring for any future attempts to violate its policies. This action follows a report by Microsoft in August, which detailed Storm-2035’s attempts to engage US voter groups on opposite ends of the political spectrum with divisive messaging on various contentious issues.
Earlier this year, OpenAI disrupted five other covert influence operations that sought to misuse its models for deceptive activities online.
