Autonomous System

Autonomous System refers to an AI-enabled system capable of performing tasks, making decisions, and taking actions in the physical or digital world with minimal or no human intervention. These systems perceive their environment, process information, and act based on their programming and learned behaviors.

Regulatory Framework: The EU AI Act addresses autonomous systems throughout its risk-based framework. High-risk autonomous systems—particularly in critical infrastructure, transport, and safety components—face stringent requirements under Article 9 (risk management), Article 14 (human oversight), and Annex III classifications.

Key Regulatory Requirements:

• Human Oversight: Article 14 requires high-risk AI systems to have human oversight capabilities, including the ability to intervene or stop the system
• Risk Management: Autonomous systems must implement continuous risk assessment throughout their lifecycle
• Safety: Autonomous safety components (Article 6 and Annex I) are automatically classified as high-risk
• Transparency: Users must be informed of the system's autonomous capabilities and limitations

Levels of Autonomy: Autonomy exists on a spectrum from fully human-controlled to fully autonomous. Regulatory requirements often increase with the level of autonomy, particularly for decisions affecting safety, rights, or critical services.

Examples: Autonomous vehicles, automated trading systems, robotic surgery assistants, autonomous weapons systems (prohibited under certain frameworks), and self-managing infrastructure systems.

Related concepts: Semi-Autonomous System, Human Oversight, Human-in-the-Loop, Meaningful Human Control