European courts scrutinize AI responsibility in public decisions and search results

The Administrative Court of the Marche region in Italy (TAR Marche, case No. 758/2026) ruled that a public procurement decision that relied on technical documentation produced with generative artificial intelligence must be subject to strict human oversight. The court emphasized the need for a "humanity reserve" to ensure accountability and prevent unchecked AI influence on administrative outcomes.

In Germany, the Regional Court of Munich examined Google’s AI Overviews feature, which automatically generates summaries of web content and displays them prominently in search results. The court found Google directly responsible for inaccurate statements that linked two German publishing firms to alleged fraud, concluding that generative AI output constitutes an original assertion attributable to the service operator rather than merely a conduit for third‑party information. The ruling highlights the legal distinction between traditional search links and AI‑generated content, underscoring emerging liability issues for tech platforms.

Both decisions signal a broader European shift toward tighter legal scrutiny of AI applications in public administration and online information services, stressing the importance of human control and clear accountability mechanisms.