Deepfake technology, which uses AI’s deep learning algorithms to simulate and forge images, audio and video, can be abused or used maliciously to endanger individual legal rights, social stability and national security, the Ministry of State Security (MSS) said in an article on Friday.
It said some “hostile foreign anti-China forces” had used this technology to generate fake videos and attempted to spread them in China to “mislead public opinion and create panic”.
The ministry added that using AI could lead to data leaks and security risks. It cited a case in which staff had used open-source AI tools that permitted users to view files without a password, allowing someone with a foreign IP address to access and download sensitive data.
The ministry also warned of algorithmic bias, where large language models might amplify discrimination if the training data reflects social biases or under-represents certain issues.
11:56
From India to China, how deepfakes are reshaping Asia politics
From India to China, how deepfakes are reshaping Asia politics
To combat these risks, the ministry said users should be cautious when granting access permission, avoid using web-based AI to process sensitive data, refrain from saving payment passwords and disable data sharing and cloud storage. They should also monitor their “digital footprints”, periodically clear AI chat histories, update antivirus software and be alert when asked to provide personal identification and bank account details.
