U.S. $1bn AI-compute partnership (DoE × AMD). The U.S. Department of Energy launched a $1 billion public-private partnership to build next-gen supercomputing capacity for AI research, raising governance questions on access, export-control compliance, and provenance of training data (Reuters).
UK “AI Growth Lab” — call for evidence (open). The government is soliciting submissions to shape a pro-innovation regulatory sandbox for AI adoption; responses due 2 Jan 2026. This sets expectations for measurable benefits and audit-grade documentation from participants (GOV.UK – PDF).
Regulation
EDPS coordination on AI enforcement (mid-Oct signal). The EDPS detailed outcomes of the 2nd AI Act Correspondents Network - ongoing cross-authority work to harmonise implementation and transparency expectations for automated decision-making (EDPS note).
NIST “Zero Draft” documentation track (context). NIST’s outline for a standard on dataset/model documentation continues to serve as the reference for procurement- and audit-ready artefacts while the first full draft is prepared (NIST outline (PDF)).
Events
Artificial Intelligence and Copyright Conference — Lisbon, 6 Nov 2025, hosted by ICC Portugal; talk by Eleonora Rosati. Event • Programme (PDF).
FORUM Institut — forthcoming IP/AI events: Mastering Patent Portfolio Management in the Age of AI (20 Nov 2025) • Munich Winter Course on US & EU Patent Licensing (4–5 Dec 2025) • How to Protect and Manage Trade Secret Assets (19 Mar 2026). Overview | Munich Course | Trade Secrets.
Academia
Hartzog — “Against AI Half Measures” (SSRN): argues for substantive, enforceable duties beyond principles—useful for mapping liability hooks in AI Act/UK frameworks (SSRN).
Katz — “Law, Justice, and Artificial Intelligence” (SSRN) compares decisions of modern LLMs with human baselines; implications for explainability, contestability, and due-process-style safeguards in deployment (SSRN).
Business
Compute funding → compliance obligations. The DoE–AMD build-out will push recipients toward export-control compliance, data-lineage documentation, and model-risk controls embedded in access agreements (see News).
Procurement pressure. The UK call for evidence foreshadows sandboxes where measurement and audit trails become entry conditions for public-sector or regulated-sector pilots (see Regulation/News).
Adoption of AI
Evidence first. NIST’s documentation track and EDPS coordination indicate that logs, datasheets, evaluations, and provenance are moving from optional to mandatory artefacts for deployers.
Takeaway
Infrastructure spending and regulatory sandboxes are converging on a single demand: prove it. Whether you’re training on public supercomputers or piloting in the public sector, your compliance posture will be judged by traceability, documentation, and safety-by-design and courts are watching.
Sources: Reuters, GOV.UK, EDPS, NIST, ICC Portugal, FORUM Institut