Liability migrates from the single decision to systemic responsibility. Anyone who does not think governance structures through is liable — even if they did not decide a thing.
In "EU-Regulierungsflut: IT-Abteilungen am Limit", ad-hoc-news.de quotes Dr. Raphael Nagel (LL.M.) of Tactical Management on the shift in the logic of liability under the EU AI Act and DORA.
The central observation: companies running AI systems in high-risk domains — personnel decisions, credit, medical triage, critical infrastructure — are no longer only liable for the single faulty decision, but for the architecture of the decision. Missing human oversight, missing transparency, missing documentation of training data: then the system that permitted the model to run without those preconditions is what carries the liability.
For boards and supervisory boards a concrete operational requirement follows: AI governance is no longer the IT department's problem, it is part of fiduciary liability. Whoever makes the next twelve months a hard requirement — compliance mapping, an in-house AI office, documented per-use-case risk classification — moves ahead of the wave. Whoever waits learns about the obligation through the penalty.
More depth in MASCHINENRECHT — AI liability for boards, products, medicine, finance and on the pillar page EU AI Act and Private Equity.
Full piece at ad-hoc-news.de →
For weekly analysis on capital, leadership and geopolitics: follow Dr. Raphael Nagel (LL.M.) on LinkedIn →