Author: Xiao Yan
Edit: Caiyun
On December 19, Douyin officially released an important announcement, announcing that it would crack down on six new types of violations, including AIGC. The so-called AIGC refers to generative artificial intelligence, which is gaining momentum at the moment. The reason why we want to crack down on AIGC is actually for the sake of the pointCrack down on new types of number raising
Douyin aims to crack down on the black and gray industry, and AIGC fraud has become an unavoidable "roadblock".
In fact, the main purpose of Douyin's heavy attack this time is to crack down on the illegal behavior of "raising fans and raising accounts" in the black and gray industry. Because Douyin was found in the daily inspection,Many black and gray industry organizations are using AIGC to falsify, steal and edit other people's live content, publish training courses in the official name, etc., and publish low-quality illegal content in batches, and try to achieve the purpose of quickly increasing fans and raising accounts.
Among them,AIGC technology is used to create avatars and impersonate real people for live streamingIt is one of the most typical and common ones. If you have the habit of brushing short **, you must have swiped the content of virtual people live streaming or talking. They either imitate the real anchor to bring goods virtually, or talk about emotional chicken soup, knowledge popularization, financial management and other aspects there.
Compared with real anchors who actually carry out live broadcasts or publish content, it is much easier to use virtual humans to complete these tasks. Just imagine, it is difficult for a real anchor to conduct a live broadcast with a duration of 8-10 hours, which is unimaginably high-intensity, but there is no problem that virtual humans cannot adapt to high-intensity work. Even if you don't do traffic delivery, it is enough to extend the live broadcast timeRaise the number to increase the number of fans, and carry out crowd portraits for fans, so as to obtain more accurate new customers。Even if the live broadcast performance of virtual people is not ideal for real people, even if there are only dozens or dozens of people in the live broadcast room of virtual people, there is still no shortage of payment orders and comments in the comment area.
Douyin once took action to plug the loophole of AIGC: it is clearly marked, and the consequences are at your own risk.
In fact, this is not the first time that Douyin has put forward specifications and requirements for AIGC.
As early as May 9 this year, Douyin officially released a message to put forward platform specifications and industry initiatives for AIGC for the first time. Knocking on the key points, there are two aspects that are very important and have attracted special attention. The first is the labeling of AIGC content, and the second is the registration and authentication of virtual humans.
First of all, let's talk about the annotation of AIGC, which is simply to "put a watermark".
Douyin clearly stated in the relevant specificationsDeep synthesis services, including AI writing, AI text editing, and AI speech synthesis, that may cause confusion, must be clearly watermarked and labeled。Whether you are using ChatGPT, Midjourney, which is popular this year, text and ** generation products such as Wenxin Yige and Wenxin Yiyan in China, or speech generation products such as Moyin Workshop, they are all in the category of "marking". For the identification method, to put it simply, it is necessary to mark the word "AI generated" in the "upper left corner" of all kinds of ** and ** files, as well as the name of the generation tool used and the company name watermark.
Most service providers are optimistic about the "watermark" required by Douyin, and generally feel that adding watermarks will not affect the spread of content, but can achieve the purpose of ensuring authenticity. Many people believe that "this is the right decision" and that it will "become the norm" in the future.
Compared with the identification of AIGC content, the situation of virtual humans is much more complicated.
At that time, Douyin made it clear in the announcement that it supported the development of virtual humans, but virtual humans needed to be authenticated, and completely unmanned AI live broadcasts were prohibited. According to the use, there are two types of virtual humans on Douyin, one is the "entertainment" virtual anchor;One is the functional virtual person, the most intuitive are those who need to complete the goal of live streaming.
However, users don't buy too much about the behavior of virtual human live streaming. **Compared with the live anchor, the live broadcast content of virtual humans is of poor quality, the language is wooden, and there is nothing new to speak of. For the majority of merchants, using virtual humans to bring goods is not an option to sign up, and virtual humans are not only difficult to complete KPIs and achieve revenue generation, but also easy to be sentenced to violations by the platform.
That's exactly how things went. After a lapse of half a year, Douyin officially raised the threshold and requirements, and directly put forward the requirement of "prohibiting AIGC fraud", showing a clear card to AIGC. It can be regarded as further clarifying the official position and attributes of Douyin.
Douyin took the lead in thundering and may become the vane of the industry.
ChatGPT has opened a new door to artificial intelligence, allowing us to see the power of artificial intelligence, and at the same time, the corresponding Pandora's box has been opened.
With the continuous maturity of technology, AIGC has gradually moved from the concept stage to the actual implementation stage, and disputes and risks have emerged. For some time now, there have been frequent cases of AI-generated content being fake, data breaches, and infringements, which further amplify the dangers that AI can pose to society.
Today's AIGC is widely used in the short track, and creators, anchors, users, merchants, advertisers and other platform ecological participants will all involve AIGC without exception. This is also an important reason why Douyin took the lead and introduced corresponding industry norms.
There is always a relationship between regulation and innovative technology and innovation models. There will be technology first, then there will be risks, and then there will be the need for standardization. With regulation, the development of the industry will develop in a more favorable direction.
Of course, the introduction of the norm does not mean that it will be done once and for all. In the process of enforcing regulation, we still have practical problems. For example, how to define specific responsibilities?How do I collect evidence?But in any case, the regulation of artificial intelligence is on the agenda, which is very gratifying.