Longer-form analysis on AI governance — what is actually changing, what it requires of an organization, and what remains unresolved. Sourced, dated, and revised in public when it turns out to be wrong.
The register took real fact-checking, and this will take the same. Rather than publish quickly and correct later, nothing appears here until it meets the standard below. If you want to see how that standard reads in practice, the register is six entries long and available now.
Regulations, judgments, official disclosures and published research — read, not summarised from coverage. Where reporting originates a finding, the reporting is named.
No illustrative figures presented as real, no composite examples presented as cases, no client work presented as anything. Any illustrative material is labelled illustrative on the page.
Where something is contested, unresolved or simply not known, the piece says so rather than choosing the reading that makes the argument tidier.
Every piece carries a date. When the underlying position changes, the piece is revised and the revision is recorded rather than quietly overwritten.
A governance firm that publishes an unsourced claim has demonstrated the exact failure it sells the remedy for. That is the reason this page is empty rather than full.
Six recorded governance failures, each traced to a primary source, each naming the control that failed, each ending in one question worth asking of your own organization.