Skip to content

Compliance & Governance

EU AI Act Compliance Statement

How empirical approaches traceability, risk controls, and human oversight in AI engineering.

Last updated:3 October 2026

Engineering approach

We design AI-enabled systems around their intended purpose, deployment context, and risk. Our work can include data provenance, evaluation criteria, traceable system behaviour, logging, access controls, human oversight, incident handling, and technical documentation.

Risk-sensitive controls

  • Classify the system and clarify the roles of provider, deployer, importer, or distributor before defining obligations.
  • Document data sources, model and component boundaries, evaluation evidence, known limitations, and change history.
  • Apply proportionate monitoring, human review, security, and fallback mechanisms.
  • Avoid presenting probabilistic model output as deterministic fact where the distinction matters.

No blanket certification claim

Compliance depends on the specific system, role, use case, risk category, and applicable transition date. This statement describes our engineering approach; it is not legal advice, a conformity assessment, or a claim that every system or engagement is automatically compliant.

Questions about this statement can be sent tooffice@empirist.com.