Influence of existing standards on EU AI Act accuracy terminology

Determine whether terminologies adopting existing standards influence the notion of accuracy in the EU AI Act's standard-setting process.

Background

The EU AI Act leaves accuracy undefined and relies substantially on harmonized standards and standard-setting organizations to specify compliance requirements for high-risk AI systems. The paper emphasizes that technical terminology may acquire different meanings when incorporated into legal and regulatory frameworks.

The authors explicitly leave unresolved whether pre-existing standard terminology will shape the regulatory meaning of accuracy during the EU AI Act's standardization process. Resolving this question would clarify how external technical standards participate in translating machine-learning concepts into legal compliance criteria.

References

Near term, it is relevant to study whether terminologies adopting existing standards influence the notion of accuracy in the EU AI act's standard setting process.

— Law of Large Numbers: Accuracy as Statistical Measure for AI Compliance and Competition  (2608.31018 - Derr et al., 31 Aug 2026) in Section "Lessons Learned -- A Satellite Perspective"