Skip to content

Regulations

One AI System Answers to Many Regimes

The EU AI Act sits at the centre. Data protection, cybersecurity, operational resilience and product safety law apply alongside it, with a growing number of regimes outside the EU. Each was written separately. We map every one down to the single obligation, so one body of evidence answers all of them.

Why this is hard

The Regimes Were Written for Different Purposes

Take one high-risk decision in a bank. The AI Act treats it as a high-risk system. The GDPR treats it as automated processing of personal data. NIS2 and DORA treat it as a resilience matter, and the Product Liability Directive as a possible defect. One incident can start reporting duties under three of them, each with its own deadline.

Different bodies wrote these rules for different purposes. Meeting one does not meet the others, and sometimes two of them conflict. A checklist that ties a control to a named act cannot settle that. It takes a map of each obligation.

EU AI Act
A high-risk system
GDPR
Automated processing of personal data
NIS2 and DORA
A resilience matter
Product liability
A possible defect

The regimes

The Regimes That Reach the Most Systems

Each has its own page: what it governs, who owes what, the timeline, the penalties, and where it meets the AI Act.

When AI Software Becomes a Product

From 9 December 2026, software is a product under EU liability law, and that includes AI. Liability becomes strict and no-fault. A defect can be presumed where a product breaks a safety rule meant to prevent the harm. The technical documentation you keep for the regulator becomes evidence you must disclose in court. Complete, your governance file is your defence. With a gap, it is the presumption against you.

Your next step

Which of the Six Apply to You?