A working paper published April 7, 2026 by Nannini et al.—"AI Agents Under EU Law: A Compliance Architecture for AI Providers"—is the most important regulatory document in the agentic AI governance space this year. Not because it proposes a solution. Because it admits, in precise regulatory language, that the solution does not yet exist. Two findings in the paper warrant careful attention. The first is Conclusion 8. The second is Observation 10. Together they constitute the first official regulatory acknowledgment of a substrate governance gap that has been architecturally visible—and documented—for some time.
Narnaiezzsshaa Truong (Tue,) studied this question.