NextAI+ Praxis--:----UTC
Column · B·Weekly, since Q4 / 2024·67 entries to date

AI Governance
Weekly

The new rules, in plain language. EU, UK, Singapore, US state-level, mainland China — what changed, who has to act, by when. We do not summarise statutes; we read them the way a compliance officer at a mid-sized firm reads them, and report what is enforceable on Monday morning.

7 entries · March — May 2026

Just filed№ 067

AI governance’s “East–West temperature gap”: Western compliance recalibration meets China’s strong-enforcement stance on agentic systems.

Two governance climates are diverging in real time. We read this week’s Brussels softening alongside Beijing’s MIIT enforcement bulletin, and chart the practical asymmetry for any company that ships an agent across both jurisdictions.

Argument№ 066

The “hardening” moment of algorithmic governance: from ethical principle to technical enforcement.

Principles are out of fashion. This week saw three regulators publish technical conformance tests — audit logs, model cards, red-team transcripts — that look less like policy and more like product specs. We read what is now machine-checkable.

Filing№ 065

Anthropomorphic-interaction red lines, log evidentiarisation, and cross-border data relay.

Three small clauses, one big shift. The week’s filings from Singapore, Korea and the EU converge on the same idea: an interaction log is now evidence, and the rules for moving that evidence across borders are no longer optional.

Bulletin№ 064

The compliance clock starts: EU “hard red lines” close in as China’s “ethics-first” rules become mandatory.

Two clocks, one window. Brussels’ high-risk obligations bite in August; Beijing’s revised algorithmic recommendations rules became enforceable last week. We line up the calendars side by side and mark the dates that change a board agenda.

Long read№ 063

2026 governance new normal: multi-country regulators tighten requirements; corporate compliance pivots to “governance capacity”.

The job description has changed. A compliance officer in 2026 does less interpretation and more orchestration: of evidence, of cross-team review, of model-lifecycle artefacts. We sit with three firms who quietly rebuilt the function this quarter.

Argument№ 062

Sovereign competition, federal centralisation, and the liability boundary of agentic AI.

Who is liable when an agent acts on your behalf — and acts wrongly? We read three sovereign-AI procurement notes and the first US-federal-agency direct-liability guidance, and trace the boundary that is being drawn in ink, not pencil.

Opening salvo · Vol. 1№ 061

2026 AI strategy landscape: the return to business beyond technological hallucination.

The first entry of this year’s volume. The deck-shaped AI strategy is going out of fashion, and a quieter species of strategy — one written by an operations lead and a counsel together — is replacing it. We list the seven signs we are watching.

Column · A
The companion column

Enterprise AI Deployment Signals

A close, weekly reading of how large enterprises actually wire AI into the business. Architecture, vendor choice, where the money lands. Filed every Thursday.

Read column A

One quarterly digest, no weekly drip.

If you want both columns delivered together, four times a year, in one quiet email — leave an address. Otherwise just bookmark this page.