Weekly Quality Report丨The image of AI doctor sells medicines. Who will bear the responsibility?
At present, advertisements for various products can be seen everywhere on online platforms. Many promoters claim in their videos that their products are “effective through personal testing.” However, upon closer inspection, many of the people featured are not real natural people, but digital virtual people generated by AI. How can a digital person without sentience have a “real experience”? Are you qualified to endorse products?
Not long ago, the producer of an AI short drama released a product placement commercial. The AI digital human declared in the first person: "I wore my contact lenses for a day and they were very comfortable." Related product links were also attached to the video. As soon as the advertisement came out, a large number of netizens immediately questioned why an avatar without physiological perception could recommend to consumers a cosmetic contact lens product that directly contacts the cornea.
On August 3, “AI short drama heroine’s 60-second advertising quotation is 258,000 yuan” was on the hot search; on August 6, the advertising video with related content was removed from the shelves by the producer.
Li Huaisheng, director of the Internet Law Institute of China University of Political Science and Law: If digital people express a unique and personalized experience that only real people can have in advertising, it may be suspected of false advertising. For example, a digital person praises a certain food that I eat very well, etc. This is obviously a kind of anthropomorphic false propaganda.
According to the reporter’s understanding, cosmetic contact lenses are decorative colored flat contact lenses. They are managed as Class III medical devices in my country and fall into the same regulatory risk level as implantable pacemakers, artificial joints, and vascular stents. Article 16 of the "Advertising Law of the People's Republic of China" clearly stipulates that advertisements for medical treatment, drugs, and medical devices shall not use advertising spokespersons to make recommendations or certifications. Experts point out that the key to whether a digital person is qualified for endorsement lies in whether there is a real person behind him or her to back up the responsibility.
Zhi Zhenfeng, a researcher at the Institute of Law of the Chinese Academy of Social Sciences: It is completely fictitious. We call it a native AI digital human. It cannot be endorsed because it does not meet the legal requirements of being able to fully try out the product and have a real experience.
However, reporters have noticed that it is very common for AI digital people to "endorse" advertisements on online platforms. Reporters browsed multiple online platforms and found that a large number of AI digital people who appeared as "medical workers" and "field experts" "endorsed" to promote related drugs and health products for eye diseases, gray hair, etc.
The reporter’s investigation found that in many product promotion videos, those who play the role of promotion are native AI digital people. During the promotion process, expressions such as “root removal” are frequently used to exaggerate the efficacy. Some videos also show so-called “scientific experiment” scenes.
According to experts, some virtual digital people deliberately imitate the image of medical staff to promote related products, which brings risks that far exceed the efficacy of the product.
Li Xu, associate professor at the School of Life Sciences, University of Science and Technology of China: The most dangerous thing about these digital doctors is not that they are selling a product that may not be effective, but more importantly, they turn artificial intelligence technology into a negative trust amplifier. First of all, it greatly lowers the threshold for impersonating a professional authority, especially allowing consumers who do not know much about AI technology and whose health literacy is not high enough to mistakenly think that they are hearing formal medical advice.
Experts believe that what is even more dangerous is that these AI digital people impersonating medical workers may cause some patients who do not know the truth to give up regular treatment.
Li Xu, associate professor at the School of Life Sciences, University of Science and Technology of China: For cardiovascular patients, this type of advertising is actually more dangerous. If the patient believes in the so-called magic medicine that dissolves blood vessel waste and cuts off the blood vessels as soon as it is taken, and stops taking the medicine he should be taking without authorization, he may push himself back to a high-risk state.
just a few minutes
can generate a video of an AI digital person bringing goods
The image of an AI digital human in a white coat has misled many consumers. So, how are these seemingly professional and authoritative digital advertising videos produced? During the investigation, the reporter found that to make such a video now, you only need to upload a photo, and in just a few minutes, a video of an AI digital person carrying goods can be generated.
The reporter searched the mobile app store and found that a large number of apps claim to be able to "generate AI videos with one click and reproduce popular products", and some charge only 48 yuan a week. In the case display, there are not only AI digital people carrying goods for various products, but also videos of restaurant visits.
The reporter downloaded and tried this App, tried to take a photo and upload it, and selected "Video with goods", and the software immediately provided a one-click generation function. After turning on one-click video generation, an 8-second video of digital people bringing goods was completed in just a few minutes.
AI digital person carries fruits: Sisters, look what good thing I have discovered. The skin of this crown pear is so smooth with small spots. It feels heavy in the hand and must be very moist.
When the reporter further mentioned in the prompt: Generate a medical worker to carry a plaster. As a result, within a short period of time, a digital person wearing a medical uniform began to shout at the plaster.
AI digital human-delivered plaster: Sisters, look what good thing I have discovered. This plaster does not look small, but the effect is very good. This plaster is not expensive, but you need to keep it at home. It is really suitable for daily use, so hurry up and try it.
Reporters found that the core selling point of such apps is “low threshold, batch generation”. As long as users purchase the service, they can batch produce a large number of digital videos of people bringing goods.
The Advertising Law clearly stipulates that advertisements for drugs and medical devices are prohibited from using spokesperson recommendation certificates, and the images of medical personnel are not allowed to be used in commercial advertisements. However, a reporter's investigation found that there are no corresponding restrictions in the algorithms of some generating software.
Pan Jiming, executive member of the Computer Security Committee of the China Computer Society: After a good technology is invented, it can be used to do many good things, and it can also be used and abused. Therefore, I think it is more of a human problem and needs to be strictly controlled on how users can use it more reasonably and safely.
There is something wrong with the products promoted by AI digital humans
Who should bear the legal responsibility?
The reporter’s investigation found that some digital human advertisements not only do not have AI logos, but some generation software even provides the function of “removing watermarks”. So, with the “AI generated” logo, can we be exempted from corresponding legal liability? If something goes wrong with the products promoted by the AI digital human, who should bear the legal responsibility?
In recent years, regulatory measures for AI digital humans have been introduced one after another, but the reporter's investigation found that some practitioners are lucky. Some do not mark the AI-generated logo, some find ways to erase the logo, and some think that as long as they are labeled "AI-generated", they can be exempted from corresponding responsibilities.
Zhi Zhenfeng, a researcher at the Institute of Law of the Chinese Academy of Social Sciences: If the advertising and marketing invested by the AI digital human is marked, then the responsibilities that should be borne if it is not marked will be exempted. However, whether the advertisement is real and whether the goods in the advertisement are fake or shoddy, these responsibilities cannot be exempted.
Digital people cannot bear legal responsibility. If something goes wrong, the responsibility will fall on the real person behind the screen. Experts point out that accountability requires clarifying the responsibilities of each actual participant behind the advertisement.
Li Huaisheng, Director of the Internet Law Institute of China University of Political Science and Law: First of all, advertising is mainly responsible for the authenticity of the content. Secondly, advertising publishers, including relevant endorsement agencies and relevant agencies that produce AI digital humans, must bear the responsibility for content review. In fact, the AI digital person itself cannot be a liability for evading legal punishment. In the process of pursuing accountability, we must also penetrate the actual subject behind the advertisement.
Experts believe that governance cannot just rely on the platform to "remove illegal videos", but must be upgraded to collaborative governance covering the entire communication and business chain. Regulatory authorities need to truly link up the chain of responsibility. Whoever created the digital person, who wrote the script, who operated the account, who purchased the traffic, who sold the product, and who ultimately made the profit should all be traceable.
Li Xu, associate professor at the School of Life Sciences, University of Science and Technology of China: Platforms need to guard the first gate. Whenever commercial health content such as disease treatment, drugs, health products, and medical devices are involved, especially those digital people who appear in professional capacities such as doctors, pharmacists, and professors, they should conduct more stringent identity authorization and qualification verification. Absolute medical commitments such as radical cure, eradication, discontinuation of medication, and wrapping up should directly trigger high-risk audits.
False advertising by AI digital humans is impacting the three ethical and legal bottom lines of "authentic content, authentic experience, and authentic identity" in commercial promotion.
Various online platforms must shoulder their main control responsibilities, upgrade intelligent risk control and semantic recognition models, and actively identify and move forward the illegally generated content that impersonates professionals and fake user experiences. At the same time, regulatory authorities must strengthen the mandatory labeling of AI-generated content, pre-advertising review, and rapid disposal of illegal content to clarify the legal boundaries of AI advertising and ensure that digital human advertising operates within the rule of law.
