Standard of care
The level of prudence and care expected of an actor, the violation of which characterizes the fault or defect grounding liability.
On February 11, 2025, the European Commission listed in its work program the withdrawal of the proposed AI Liability Directive, tabled on September 28, 2022. That leaves Regulation (EU) 2024/1689, the AI Act, to set the standard of care for AI actors indirectly. The standard of care is the prudent and careful conduct expected of an actor; breach it, and liability can be established. With no dedicated instrument, the regulation's obligations tend to become the reference: documented risk management, data quality, traceability and human oversight. Compliance weighs in assessing the safety a person is entitled to expect, without creating a presumption of it. Breach of a mandatory safety requirement, on the other hand, triggers a presumption of defectiveness under Directive (EU) 2024/2853 on defective products, or supplies the fault under national laws. Compliance and liability, conceived as two distinct pillars, thus merge, and the standard of care also becomes the natural criterion for insurance underwriting.
Textbook case: an AI deployer who neglects the human oversight required by the AI Act breaches the standard of care, which may suffice to characterize a fault before a national court.
Follow the research
One letter a week. First, the week's underwriting note, when one is published: a dated, argued decision, signed by Paul Bernardski. Then what the observatory published on cyber risk, political risk and insurance. No account, no tracking.
No spam, one-click unsubscribe, no data ever sold.