Leading AI safety company OpenAI has taken action against an Iranian organization that was utilizing its conversational model ChatGPT to create politically charged content focused on topics surrounding the US elections.
Through thorough investigation, OpenAI discovered that the operation, code named Storm-2035, had established multiple social media accounts and websites which they were using to share articles and posts generated through their ChatGPT accounts. The content zeroed in on issues like the presidential candidates, conflicts in Gaza and Israel, and other sensitive geopolitical subjects.
It appears the operation was hoping to influence online discourse related to the US elections scheduled for November through spreading this politically motivated content. However, OpenAI notes that majority of the social posts received very little engagement, indicating the effort did not achieve significant reach or impact audiences widely.
As a precaution, OpenAI has now blocked all accounts linked to Storm-2035 from accessing their services going forward. They will continue monitoring for any attempts to misuse their models or violate the usage policies. The dismantling of this operation demonstrates OpenAI's commitment to ensuring its innovative technologies are not exploited for harmful aims such as covert online influence campaigns.
The timely action against this Iranian group shows the responsibilities of AI developers to protect against the misappropriation of their models. As ChatGPT and other conversational technologies continue advancing, companies must stay vigilant in disrupting any activities that threaten election integrity or undermine civic processes through sophisticated digital manipulation.
