AI Dictionary › Regulation
Approccio basato sul rischio
The risk-based approach is the methodological principle whereby the regulatory obligations imposed on an AI system are proportionate to the level of risk that system may pose to people, rather than being uniform across all systems regardless of their impact. It is the cornerstone principle underpinning the EU AI Act and also guides other international frameworks such as the NIST AI RMF.
In practice, the approach works through classification: systems are placed into risk tiers (for example unacceptable, high, limited, minimal) based on criteria such as the sector of application, the system's decision-making autonomy, and its potential impact on fundamental rights, health or safety. Each tier carries different obligations: minimal-risk systems face almost no constraints, while high-risk systems must meet strict requirements on risk management, data quality, transparency and human oversight.
For companies, adopting a risk-based approach means first mapping their AI system portfolio and classifying it correctly: a misclassification can lead to missed obligations or, conversely, unnecessary compliance burden. It is an exercise requiring both technical and legal expertise, often carried out with specialized consultants or dedicated risk-assessment tools.
The concept has roots in European product regulation, where similar logics have long applied to medical devices, toys and industrial machinery. It was later adapted to artificial intelligence as an alternative to a blanket ban or uniform regulation, allowing low-impact applications to proceed unhindered while concentrating controls where potential harm is greatest.
From our network
AGORÀ Intelligence: Enterprise AI Governance Platform
Govern AI at scale: policies, adoption and measurable results on your data. Built for boards and C-suite.
Visit agora-intelligence.com →From the Agora Intelligence blog
📱 Download the Android app (beta) iOS coming soon
Say what you mean. Get what you need.
Grace Certified, the AI coach that trains and certifies your prompt engineering, by Agora Intelligence.