Accounts Comprising at Least 686, Organized Activities
"Use of Awkward Expressions Rarely Used by People"
A study has revealed that hundreds of fake accounts using artificial intelligence (AI) on the social networking service (SNS) platform X (formerly Twitter) have been systematically impersonating people and spreading posts supporting former U.S. President Donald Trump and the Republican Party.
On the 17th (local time), NBC News cited a recent research report published by Clemson University researchers to report this. The researchers identified an organized network consisting of at least 686 X accounts. These accounts have posted more than 130,000 posts on X since January. The posts mainly concern four recent U.S. Senate primaries and presidential candidate primaries. They also express support for former President Trump and the Republican Party.
The Clemson University researchers noted that these accounts used large language model (LLM) technology, which is employed in AI chatbots like ChatGPT, to interact with other users. According to the report, these accounts consistently showed patterns such as setting profile pictures with memes popular among far-right supporters, crosses, or the U.S. flag, and regularly replying to other people’s posts on political issues.
NBC News reported, "These accounts have almost no followers," and described their method as "replying to posts that have already attracted attention among people." The Clemson researchers confirmed this by detecting traces of a kind of 'error' indicating that posts were written using AI chatbots like ChatGPT. Some accounts used awkward expressions rarely used by people.
The researchers viewed the operators or financiers behind this organized network of AI accounts as unclear. They pointed out that since the content of the posts focuses on U.S. domestic politics and lacks clear connections to foreign countries, it suggests the possibility of a political operation by internal U.S. forces rather than a foreign government.
Recently in the U.S., as the November election approaches and AI technology spreads rapidly, warnings about attempts to influence election polls using AI have been consistently raised.
NBC News stated, "The potential existence of domestic operational forces using AI adds another concern to the rapidly advancing AI technology and the resulting chaotic information landscape." Most of the AI accounts identified in the Clemson University study were confirmed to have been deleted after NBC requested a statement from X. X did not respond to NBC News’ request for comment on this matter.
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
