China’s Cyberspace Administration has issued draft new rules aimed at strengthening oversight of artificial intelligence services designed to mimic human personalities and interact emotionally with users. These new rules require service providers to ensure the ethical, security and transparency aspects of their services.
In a statement published on its website, the Chinese Authority clarified that users must be informed that they are interacting with artificial intelligence upon logging into the service, once every two hours, or when any signs of excessive reliance on this technology appear.
This step reflects Beijing’s efforts to control the rapid expansion of artificial intelligence services available to the public, by imposing stricter standards related to safety and ethics.
The proposed rules will apply to artificial intelligence products and services offered to consumers in China, which display human personality traits, thinking patterns, and simulated communication styles, and interact with users emotionally through texts, images, audio, video, or any other means.
The new draft defines a regulatory approach that obliges service providers to warn users of overuse, and to intervene when signs of addiction appear in users.
According to this proposal, service providers will be responsible for safety throughout the product life cycle, and will have to put in place systems for reviewing algorithms, data security, and protecting personal information.
These measures also define red lines for content and behavior, and stipulate that service providers must not create content that threatens “national security” or spreads “rumors” or promotes “violence” or “obscenity”.
China attaches great importance to artificial intelligence as a strategic industry, and invests heavily in new technologies to support economic growth and enhance global competitiveness. At the same time, the authorities are keen to maintain governance to ensure security and social stability.
According to the draft proposals, service providers must conduct a security assessment and submit a report to the “Provincial Cyberspace Administration” if any human-simulating artificial intelligence features are launched. A report will also be required for services that obtain “one million registered users” or “100,000 monthly active users”.
