By Jill Hubbard Bowman
This blog will explore the European Union’s new Artificial Intelligence Act (AI Act), which was recently approved by 27 member states, as it is implemented over the next year. The council of the EU recently released the text of the provisional agreement which includes the key terms for understanding the scope of regulated AI technology. A critical definition in the 272-page agreement is “AI system” because it defines, with some exceptions, what type of technology will fall within the scope of the Regulation. (Note the "AI System" definition will have future revisions as the EU AI Act is finalized and implemented.)
AI System
“‘AI system‘ is a machine-based system designed to operate with varying levels of autonomy and that may exhibit adaptiveness after deployment and that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments;”
Recital 6 on page 16 of the provisional agreement details the characteristics.
Essentially, an AI system:
Inference
“[I]nference refers to the process of obtaining the outputs, such as predictions, content, recommendations, or decisions, which can influence physical and virtual environments and to a capability of AI systems to derive models and/or algorithms from inputs/data.”
Inference techniques include machine learning and logic approaches based on encoded knowledge and symbolic representation.
What’s not covered by the AI Act?
Be careful. The devil will be in the details. Stay tuned for the implementing laws and later interpretations to really understand what is and isn’t regulated, but the general exemptions have been outlined in the AI Act provisional agreement.
The gist of some of what’s not covered:
Another post will define high-risk AI systems which is another key term for the AI Act.
AI Act Text
https://data.consilium.europa.eu/doc/document/ST-5662-2024-INIT/en/pdf
[i] AI systems should be distinguished from “traditional software systems or programming approaches, which are based on the rules defined solely by natural persons to automatically execute operations.” Recital 6
[ii] “This Regulation shall not apply to AI systems and models, including their output, specifically developed and put into service for the sole purpose of scientific research and development” (5a. p.96)
[iii] “This Regulation shall not apply to any research, testing and development activity regarding AI systems or models prior to being placed on the market or put into service; . . .The testing in real world conditions shall not be covered by this exemption.” (5b. p.97)
[iv] “This Regulation shall not apply to obligations of deployers who are national persons using AI systems in the course of a purely personal non-professional activity.” (5c. p.96)
[v] “This Regulation shall not apply to AI systems if and insofar placed on the market, put into service, or used with or without modification of such systems exclusively for military, defense or national security purposes, regardless of the type of entity carrying out those activities.” “This Regulation shall not apply to AI systems which are not placed on the market or put into service in the Union, where the output in used in the Union exclusively for military, defense or national security purposes, regardless of the type of entity carrying out those activities.” (3 p.95)
[vi] “The obligations laid down in this Regulation shall not apply to AI systems released under free and open source licenses unless they are placed on the market or put into service as high-risk AI systems or an AI system that falls under Title II and Title IV.” (5g. p.96)
