Written by Tiffany Cunha, GRC & Data Protection Specialist at Palqee Technologies
Updated by Ana Carolina Teles, AI & GRC Specialist at Palqee Technologies
The European Union has introduced the EU Artificial Intelligence Act, a comprehensive law that regulates the use of AI systems in the EU. In our "Decoding AI" series: The European Union’s Take on Artificial Intelligence’, we break down everything you need to know about the law for you.
Get your free copy of Palqee’s EU AI Act Framework. Get early access to the BETA program of PAM: our AI observability solution for AI systems! |
What is considered an AI system under the EU AI Act?
As defined in the final text of the EU AI Act, an 'AI system' refers to:
a machine-based system that is 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;
Here's a breakdown of the main components of this definition and their implications:
Machine-based System: This indicates that the AI system is fundamentally based on digital or mechanical technology.
Designed to Operate with Varying Levels of Autonomy: AI systems are designed to function with minimal to significant human intervention. This range of autonomy highlights the diverse applications of AI, from semi-autonomous systems requiring frequent human oversight to fully autonomous systems that operate independently.
May Exhibit Adaptiveness After Deployment: AI systems are capable of learning or evolving their behaviour based on new data or experiences after being deployed. This adaptiveness is an important feature of many advanced AI systems, enabling them to improve their performance or adapt to changing environments over time.
Infers How to Generate Outputs Based on Input: AI systems process inputs (data) they receive to generate outputs. These outputs can be varied, such as predictions, content, recommendations, or decisions. The process involves complex data processing algorithms, including but not limited to machine learning, to infer the appropriate outputs from the given inputs.
Outputs Can Influence Physical or Virtual Environments: The definition explicitly acknowledges the impact of AI systems, stating that the outputs generated by AI systems can have tangible effects on both physical and virtual environments.
For Explicit or Implicit Objectives: AI systems are deployed for specific purposes that may be clearly defined (explicit) or inferred (implicit) based on the context of their use. This aspect ties back to the intended functionality of the AI system within its operational scope.
Do all AI systems need to comply with the EU AI Act?
The EU AI Act focuses on providers, deployers, importers, and manufacturers whose AI systems operate within or are used in the European Union. It introduces a clear, risk-based regulatory framework for what it classifies as "high-risk AI systems". These systems must adhere to requirements for trustworthy AI and undergo conformity assessments before they can enter the EU market.
However, the Act does not apply to:
AI systems used exclusively for military, defense, or national security purposes are excluded.
AI systems developed or used solely for scientific research and development are also exempt.
AI systems that are neither placed on the market nor put into service within the EU and are used solely for military purposes outside the Union are not covered.
AI systems used by individuals for purely personal and non-professional activities
The legislation does cover AI systems used by public authorities in third countries and international organisations in the context of law enforcement and judicial cooperation with the EU, provided they ensure adequate safeguards for fundamental rights.
Comments