Malicious Use of ChatGPT by Chinese Actors
Some Chinese actors are reportedly using ChatGPT for harmful purposes, targeting political sectors in the United States and crafting social media posts aimed at gathering information on military technology, according to OpenAI.
In an operation called “Uncle Spam,” a group linked to China is said to employ ChatGPT to stir up controversial discussions around U.S. politics. They generate critical social media posts and create comments from various accounts, as outlined in a report released by the company.
OpenAI notes, “It appears this effort is less about promoting a specific ideological viewpoint and more about exploiting existing political divides.” This strategy is referred to as impact manipulation.
OpenAI intends to disrupt the activities on Meta after it was discovered that these actors were posting during normal work hours in China over several days.
Additionally, they have used ChatGPT to fabricate identities of U.S. veterans, particularly those who are critical of President Trump, and designed logos for social media pages, including a group referred to as “Veteran Justice.”
These actors also requested code from ChatGPT, which could potentially extract personal data from platforms like X and Bluesky.
While the number of these operations has increased, OpenAI mentions their overall impact has been limited, as the social media accounts typically have small followings.
Another faction of Chinese actors employed ChatGPT to produce polarized comments on issues like cuts to USAID funding and tariffs.
For instance, a comment on a TikTok video regarding USAID’s funding stated:
“$7.9 million of $7.9 million allocated to teach Sri Lankan journalists to avoid binary gender languages. Is this the best way to use the Development Fund?”
Some posts on X defended the reduction in USAID funding as a necessary measure to balance out tariffs.
One comment read, “Duties make imports incredibly expensive, but the government is spread thin on foreign aid. Who’s going to foot the bill?”
Another remark stated, “While tariffs are crippling us, the government spends money on ‘funding’ foreign politics.”
In a separate operation linked to China, users posing as professionals in Europe or Turkey interacted with journalists on social media, seeking information about the U.S. economy and classified documents, using ChatGPT to assist with translation.
OpenAI reports that accounts affiliated with the People’s Republic of China and certain malicious actors are being banned from the platform.
These accounts have called upon ChatGPT for support in software development and research related to U.S. military networks and government technologies.
OpenAI frequently publishes reports on malicious activities on its platform, including fake content on social media and the creation of harmful malware.
