OpenAI said on Friday it had deactivated several accounts linked to an Iranian group allegedly exploiting its ChatGPT chatbot to create content to influence the US presidential election and other issues.
The operation, known as Storm-2035, was identified as a covert Iranian influence campaign that utilized ChatGPT to produce and disseminate content on various politically sensitive topics.
According to the Microsoft-backed artificial intelligence company, these included commentary on both sides of the US elections, the Israel-Hamas conflict, Israel's participation in the Olympic Games, and issues related to politics in Venezuela.
The content was shared through social media accounts and websites designed to mimic legitimate political news outlets, aiming to sway US voters.
OpenAI reported that the operation seemed to have had minimal impact in terms of audience engagement. The majority of the identified social media posts garnered little to no interaction, with few likes, shares, or comments. Additionally, there was no significant evidence of the associated web articles being circulated on social media platforms.
Earlier in August, a Microsoft threat intelligence report highlighted Storm-2035's activities, among other things, in orchestrating influence campaigns designed to sway voters, particularly in critical swing states.
These operations also included efforts to gather intelligence on political campaigns, potentially laying the foundation for future interference. The report disclosed that Storm-2035 established covert news sites aimed at left-leaning and conservative US voters, utilizing AI to plagiarize content and propagate polarizing political narratives.
Additionally, another group, Sefid Flood, has been preparing since March for potentially extreme influence operations, including intimidation or incitement of violence against political figures, with the intent to create chaos, undermine authority, and erode confidence in the integrity of elections.
Also, in May, OpenAI published an unprecedented report revealing that it had identified and disrupted five online campaigns that sought to deceptively manipulate public opinion and influence geopolitical dynamics. These efforts were orchestrated by state actors and private entities across Russia, China, Israel, and Iran.
This development unfolds against the backdrop of an increasingly competitive race between Democratic candidate Kamala Harris and her Republican opponent Donald Trump, as they approach the pivotal November 5th presidential election.
Last week, Trump disclosed that his campaign had been compromised in a cyberattack, which he attributed to a phishing email orchestrated by an Iranian hacking group.
