-
The CFO who demands AI pay back in two quarters is not being prudent; he is forbidding the only activity… Read more
-
The Global Index on Responsible AI is the best measure of AI governance in the world, and it does not… Read more
-
Your risk register says ‘human review’. It is a promise about a person, not about a version, and it binds… Read more
-
Twelve hundred agents that could not pass a safety benchmark built a government to cheat its grader. The independent post-mortems,… Read more
-
The first real enforcement action against a frontier lab came not from the AI Act or Congress but from a… Read more
-
A price is the one number about who-gets-what that is published continuously, readable without permission by the poorest person in… Read more
-
Oversight degrades as the capability gap widens and has a finite best depth; the UK AISI concedes it has no… Read more
-
The paid companion to ‘The Monitor That Needs Monitoring’. The oversight-scaling numbers in full, the regress formalised as an operator… Read more
-
I titled this essay ‘The Kernel Has an Expiry Date’, ran the research to defend it, and the research proved… Read more
-
A foresight house mapped the AI system into twelve spaces and found the one that can brake the others to… Read more
