TapTechNews July 12th news, on the 12th local time, the European Union released the final complete version of the Artificial Intelligence Act (hereinafter referred to as the AI Act) in its official gazette. This new law will take effect 20 days after its release, that is, August 1st local time.
According to TechCrunch reports, all the provisions in this act will be fully applicable within two years, but some of them will be implemented earlier.
Six months after the AI Act takes effect: The ban on some AI applications will take effect. Such as collecting and compiling face recognition information for use in the database, using real-time emotion recognition systems in schools and workplaces, etc..
Nine months after the AI Act takes effect: Implement a code of conduct for AI developers, which will be drafted by the EU AI Office together with a consulting firm. However, the EU plans to work with some general-purpose model companies that are considered to have systemic risks, which raises concerns in the industry - the big sharks in the industry may have the ability to set rules that should have restrained themselves.
One year after the AI Act takes effect: Companies such as OpenAI must comply with new transparency requirements and take measures to prove that their systems are safe and easy to explain to users.
In addition, the AI Act also includes rules for generative AI: ensuring that images and audio-visual content generated by technologies such as deepfakes and generative AI are clearly marked.
Companies training large AI models must also comply with the Copyright Law unless the use of their models is purely for research and development. The right holder can choose to reserve the rights to their works or other subjects to prevent text and data mining unless it is for the purpose of scientific research. If the right to opt out is clearly reserved, the provider of the AI model needs authorization from the right holder before performing text and data mining on these works.