Reuters mapped the US liability landscape for rogue AI agents. This article examines the EU mirror — the withdrawn AI Liability Directive, Belgian fault law, and the evidentiary gap common to every version of the standard.

An AI incident at OpenAI and three related incidents at Anthropic show a structural mismatch: public narrative forms in hours, verified internal reconstruction takes weeks. This article examines what that mismatch means for AI Act Article 73 notifications and Product Liability Directive litigation, and what a pre-execution seal changes.

Article 9 of Directive (EU) 2024/2853 presumes a product defective if the manufacturer won't hand over technical evidence. That closes the refusal problem. It leaves untouched a narrower one: whether the evidence handed over was dated before the dispute began. 

The Digital Omnibus on AI defers the enforcement of high-risk obligations but leaves evidentiary exposure unchanged. When a system acts before the deferred deadlines, the governance state operative at T-0 remains the central question in civil, insurance, and contractual proceedings. The postponement affects regulatory temporality, not the requirement to hold an independently fixed, opposable governance record. The proof gap persists throughout the deferral window, closed only by a pre-execution cryptographic attestation architecture such as SOURCE 0.