The European Parliament of the European Union (EU) held a plenary meeting on the 14th and passed a package of artificial intelligence regulations including conversational artificial intelligence (AI) "chat GPT" and other generative artificial intelligence. The Council of 27 member states, the Council of Europe, and the European Parliament, will begin consultations on the details of the bill and strive to reach an agreement within this year.
The European Union is the world's first region to aim to fully manage AI technology, so it is attracting attention. The adoption of the bill requires approval by 27 countries and the European Parliament after the agreement between the three institutions of the European Union, and it is not expected to be implemented until after 2024. The bill requires notification of the fact that AI-generated articles and images are not generated by humans, but by AI, meaning that AI-generated works must be labeled in the future. It also stipulates that models must be designed to prevent the generation of illegal content, and AI-created works must be publicly generated, so as to eliminate the output of illegal content at the root.
In addition, the bill also prohibits police and others from using high-tech equipment for facial recognition technology in public spaces. It is also prohibited to arbitrarily collect biometric data from SNS and surveillance camera images for the purpose of creating a facial recognition database on the grounds of violating human rights and privacy. In fact, the European Commission issued the bill as early as April 21. The risks posed by the use of AI were divided into four stages and regulations were formulated, but the plan at that time barely mentioned the generation of AI. Now based on the rapid proliferation of generative AI represented by chat GPT, the European Union has added new regulations to ensure transparency in the use of generative AI.
In general, although there are many types of artificial intelligence, I think it is better to strictly restrict image generation AI and speech generation AI as soon as possible. Not only copyright, but also crimes of malicious exploitation of loopholes are likely to increase sharply in the future. Learning data reproducing the faces and voices of specific celebrities has been published on the Internet, and videos explaining their production techniques have been released one after another. A few weeks ago, it was reported that voice data generated by artificial intelligence broke through voiceprint recognition. I don't think it's too far from the report that face recognition has been breached. When technology that can easily reproduce the faces and voices of others is disclosed, who will think it is convenient to use it? The images and audio recordings used in trials can also be falsified, making it more difficult to distinguish between genuine and false.
So I think there needs to be some degree of regulation of AI. However, if "informing the generation of AI-generated articles or images is not generated by humans, but by AI-generated annotations", how can it be proved that the work is AI-generated if this regulation is violated? If you can see at a glance that something generated by artificial intelligence is illegal, and something so exquisite that it seems to be artificial, how can you judge? This is a problem