Each edition asks one jurisdiction two questions: which rules actually bind, and what happens when AI reaches a courtroom.
A thin binding layer over circulars, codes and drafts, with an observatory watching how platforms answer the one binding AI-specific duty.
No binding AI-specific rule yet. What binds is general law, written for other things. The proposed AI Governance Bill entered public consultation in July 2026.
The rules written for AI are the binding ones: NPC statutes at the base, CAC measures on top, and 2026 frontier frameworks for agents, companions and robots, with several first rulings anywhere in its courts.
A statute-led newcomer: a dedicated AI Law in force since January 2026, a synthetic-content labeling duty backed by administrative sanctions, and a Digital Code behind it.
A policy-led profile where nothing binding is AI-specific: a cabinet-approved National AI Policy sits over general cybercrime law, and the deepfake docket is rising.
No longer strategy-only: a January 2026 law writes no-sole-reliance and data duties into the statute book, while the rights questions stay untouched.
A sparse jurisdiction under military administration: content-control law does the governing, the 2019 IP laws became operational only recently, and a draft AI strategy sits unadopted.
Sparse but building: an anti-fraud law of 2026 in force, four digital bills in draft, and a completed UNESCO readiness assessment guiding the build.
A deep general digital stack with no AI-specific law: eight binding statutes an AI dispute would meet, and a national AI strategy still in formulation.
A binding AI chapter inside a Digital Code, in force February 2026: danger assessment, a high-risk class, and disclosure duties an observatory can watch.
Policy-led and outward-facing: the first national AI strategy in its region and a UN resolution of its own initiative, with no binding AI duty at home.
Policy over law by design: two national AI policies that bind no one, general statutes that do, and a confirmed founding seat in the World AI Cooperation Organization.
A digital statute book rewritten almost entirely in April 2026, threading artificial intelligence through the definitions of its cyber and telecommunications law and criminalising the worst uses of synthetic media, while asking nothing of anyone who builds or deploys a system.
17 instruments · 8 binding · 1 soft · 1 draft · 7 regional · 4 matters trackedThe region's most developed AI-governance debate and, as yet, no general AI law: a risk-based bill through the Senate, an electoral deepfake ban already enforced in court, and a real docket of decided cases.
State-led and policy-only: a 2024 national AI strategy that binds no one, general data and content law that does, and a confirmed founding seat in the World AI Cooperation Organization.
The emptiest profile in the series: no AI law, strategy or ethics code, a data-protection authority that was never created, and a confirmed WAICO founding seat signed by the state telecoms regulator.
The most drafted and least in force: a stalled AI bill, a voluntary ethics code, no data-protection statute, and a patent regime dating to 1955.
The most AI-accommodating jurisdiction in the founder set: the only patent office to name an AI as inventor, a computer-generated-works copyright provision, and a 2025 G20 presidency.
A developed digital jurisdiction with an enforced data regime, a British-model computer-generated-works copyright clause, and an AI bill now before the Senate.
An early West African data-protection adopter that leads AI with two national strategies, with patents granted regionally through OAPI.
State-led through a digitalization high-commission and a national AI strategy, with a 2025 data-protection amendment adding limits on automated decisions.
State-led through a dedicated AI institute and a 2024 data-protection proclamation; it acceded to the Paris Convention only in 2025.
Foundations built fast: a 2024 data-protection law and a national AI strategy, with patents granted regionally from OAPI headquarters in Yaounde.
At the very start of its AI-governance build: no AI law, a data authority created only in 2025, and a strategy just entering drafting.
Among the sparsest: no adopted AI strategy, a data-protection commission that was never appointed, and patents run through ARIPO.
A busy legislative pipeline with little in force yet: a data-protection bill and an AI strategy both still pending before the Assembly.
Strategy-led with an enforced data regime and a co-facilitation of the UN Global Digital Compact; its copyright law names the computer-generated work only to deny it moral rights.
No dedicated AI law and no binding AI-specific duty: AI sits inside a digital-economy decree and a data law, while the distinctive weight is regional, in the Shanghai and Eurasian blocs and a founding seat.
A dense system one signature from its first general AI law: a framework statute adopted by both chambers in July 2026 but not yet in force, above forceful general law and a genuinely adjudicated deepfake docket.
A European Union candidate that signed the Paris statement and ratified the modernized Council of Europe data convention, yet also founded the China-led organization, with a binding AI law still in drafting.
On 16 July 2026, twenty-nine states signed the founding agreement of the World Artificial Intelligence Cooperation Organization (WAICO) in Shanghai, with UN Secretary-General Antonio Guterres in attendance and Foreign Minister Wang Yi signing for China. The body had been proposed a year earlier by Premier Li Qiang at the World AI Conference, alongside a thirteen-point Global AI Governance Action Plan, and is now the first standing intergovernmental AI organization headquartered outside the Western world.
Bangladesh became the first state to announce it would join as an observer rather than as a founder, on 1 August 2026; that entry is recorded as an announced decision, since no instrument has been signed and the organization has published no observer roster.
Its design is the point. A Tsinghua-led working paper mapping WAICO against fifteen existing institutions finds:
The founding roster bears that reading out, drawn largely from states that sit outside the Western-led clubs.
This series takes that roster as its target map. WAICO has not yet published standards, model rules or an index of its own, so what its members' law actually says is the record that matters, and these profiles document it country by country, from primary sources, at reference depth. A reader researching any WAICO member should be able to come here rather than to a generic tracker. Profiles for all twenty-nine founding members are live, from primary sources at reference depth; the founding roster is complete.
Founding agreement and roster reported by Xinhua, 16 July 2026 (report ↗). Independent research and educational profiles; not affiliated with, or endorsed by, WAICO or any government.