This was a week of institutionalisation. We saw AI governance move far beyond theoretical principles and policy teams. It is officially entering the realm of statutory institutions, sector-specific operational controls and evidence capable of surviving legal scrutiny.
Here are the key takeaways from the first AIGoN Weekly Briefing:
1. A New Centre of Gravity in the Global South
Representatives of 29 countries signed an agreement in Shanghai to establish the World Artificial Intelligence Cooperation Organization (WAICO). With founding members including Brazil, Indonesia and Pakistan, it is clear that international AI governance is acquiring competing institutional centres. We cannot assume everything will converge around an EU or US model.
2. Infrastructure is Now an AI Governance Issue
New York has paused permits for new hyperscale data centres, while Australia’s new central Office of AI is preparing mandatory requirements for data centre power and water efficiency. If you are managing AI risk, you now need to collaborate with your sustainability and procurement teams. Compute is no longer just a technical dependency.
3. The Danger of "Hidden" AI Metrics in the Workplace
In the US, a court declined to halt Meta layoffs where employees allege that AI-assisted performance systems unfairly selected workers who had taken protected leave. While the claims are unproven, the lesson for HR and risk leaders is clear: you must be able to reconstruct exactly how automated metrics and human judgements interact. Simply stating that a human "formally approved" a decision will not survive legal scrutiny.
The Bottom Line
An AI inventory organised only by "models" is no longer enough. Real compliance means mapping the entire chain: the training data, the user interfaces, the infrastructure and the human reviewers.
Are your current internal guardrails ready for this shift?
Read the full, in-depth analysis and track the global policy timeline in the main edition of The AI Governance Network brief.