July 11 News.European UnionThe Commission issued a notice yesterday (July 10) launching the final version of theGeneral artificial intelligenceCode of Conduct (General-Purpose) AI Code of Practice), which helps organizations comply with regulatory standards as they develop AI.
Developed by 13 independent experts with input and suggestions from more than 1,000 stakeholders, including AI developers, academics, civil society organizations, copyright holders, and security experts, the Guidelines provide practical tools to meet regulatory standards for transparency, copyright, and safety and security.
According to the EU Guidelines, a general-purpose AI model is an AI model that can perform a wide range of tasks and can be integrated into downstream applications.
Such models are usually trained based on large and diverse datasets and are at the core of AI services such as language generation, multimodal content creation, etc. IT Home quotes the EU website, which states that the Guidelines are centered around safety and security (Safety and Security), transparency (Transparency) and copyright (Copyright) Three areas unfold.
In the Transparency section, the Code provides a sample model documentation form for companies to understand the information they need to provide in their mandatory reports.
Once the Guidelines have been formally endorsed by member States and the Commission, enterprises can sign them voluntarily in order to reduce the administrative burden and increase legal certainty.
The EU is accelerating the implementation of the Guidelines, and the governance and compliance provisions on generalized AI will be formally implemented from August 2 this year.
Under the relevant provisions of the Guidelines, AI models built before August 2 have 2 years to deploy adjustments to comply with the new regulations, while AI models built after that date have 1 year to deploy adjustments.