deepfakes they can be fun, but they also do damage. This concern reached the government of China, which started to request identifications, such as watermarks, in content made with artificial intelligence (AI). The measure aims to “safeguard national security and social public interests”.
The decision comes from the Chinese Cyberspace Administration (CAC, in English), the body responsible for overseeing the internet in the country.
In a publication this Sunday (11), the government revealed the “Regulation on the Administration of Deep Synthesis of Internet Information Services”.
In the letter, the authorities acknowledged that AI-generated content such as images, texts, videos and voice, including deepfakes, are beneficial. But technology is also used by malicious people.
“In meeting user needs and improving user experience, technology has also been used by some unscrupulous people to produce, copy, publish and disseminate illegal and harmful information, slander and belittle other people’s reputation and honor, and falsify the identity of other people,” they listed.
For the body, this attitude “affects the order of communication and the social order, harms the legitimate rights and interests of the people and endangers national security and social stability”.
And that’s where the novelty comes in.
China wants more control over deepfakes
The regulation, according to the statement, helps to prevent and resolve “security risks”.
In addition, the initiative intends to “promote the healthy development of in-depth synthetic services and improve the level of supervisory capacity”. Therefore, those responsible for deepfakes should add signs such as watermarks to avoid public confusion or misidentification.
According to the document, the requirement is intended for “services that provide functions such as intelligent dialogue, synthesized human voice, human face generation and immersive realistic scenes that generate or significantly alter the information content”.
It will also not be allowed to edit the contents to remove the identification.
The measures will come into effect on January 10, 2023.
Deepfakes are cool, but also a problem
Undoubtedly, this is an important measure and avoids some headaches. In the end, as much fun as deepfakes are🇧🇷 technology can be used to misinform and even tampering with records to incriminate someone, for example.
Of course, threats don’t just apply to this type of threat. Especially because there is a risk of editing for revenge porn videos, in addition to other embarrassing situations.
Regardless of the reason, the point is: even though it’s not easy to create well-made deepfakes, these edits can reach unsuspecting viewers.
However, we are talking about situations that, at various times, border on illegality. Therefore, even if China creates the regulation, the practical effect may be small when considering this factor. On the other hand, the Chinese government has one more legal apparatus to act in these situations:
“When major information security risks are found, cybersecurity and IT departments and relevant competent departments may, in accordance with their responsibilities, require service providers to take measures such as suspending information updates, registering user accounts or other services,” they reported. “Clarify legal responsibility for violation of regulations.”
With information: Ars Technica and The Register