A recent cybersecurity report from OpenAI reveals that even the Chinese government has employed ChatGPT for its operations. This finding emerged after OpenAI identified and shut down an account linked to the Chinese authorities, which was using the AI tool to enhance periodic status updates.
In its report titled "Disrupting Malicious Uses of AI," released on Wednesday, OpenAI dedicated a significant portion to detailing China's strategies for suppressing and intimidating critics on social media platforms owned by U.S. tech companies.
OpenAI described these efforts as extensive, resource-driven, and ongoing, involving hundreds of personnel and thousands of fabricated accounts spread across various platforms. They also noted the deployment of localized AI models along with a comprehensive set of tactics.
Targets of these operations extend beyond Chinese borders, impacting dissidents globally and even foreign officials, including the Prime Minister of Japan.
The report highlighted that Chinese operatives have engaged in activities such as fabricating U.S. county court documents to submit to social media platforms, aiming to have specific posts taken down.
Additionally, these operatives have employed hundreds of fake accounts to file fraudulent reports against dissidents, resulting in unwarranted bans and content limitations on their profiles. OpenAI indicated that many of these misleading reports included AI-generated images that mimicked screenshots from discussions.
Among the prominent accounts affected are those with substantial followings on X, including the account @whyyoutouzhele, known for sharing content from China that frequently critiques corruption and human rights violations. This account has gained significant traction, boasting over 2.1 million followers as of January.
In a recent message on X, the team behind this account urged major social media platforms like X, YouTube, and Bluesky to recognize that their content moderation systems are being exploited by the Chinese Communist Party. They stressed the importance of the tech industry acknowledging its role in enabling human rights abuses, arguing that simply claiming to be neutral toolmakers is unacceptable.
Other social platforms have noted similar trends and are taking steps to address these issues. Aaron Rodericks, who leads trust and safety at Bluesky, commented that the platform is actively working to identify and dismantle such influence operations. This includes employing specialized investigative staff and enhancing their monitoring capabilities.
Rodericks mentioned that Bluesky has recently removed several accounts that engaged in coordinated inauthentic activity that violated their community standards, in line with the findings of the OpenAI report.
Similarly, a source indicated that Meta regularly includes the activities highlighted by OpenAI in its own adversarial reporting and takes action against accounts that breach its policies.
At the time of this report, there was no immediate response from either X or China's Ministry of Foreign Affairs.