Latest writing
-
What Are We Using All This Intelligence For?
If AI is as capable as we are told, what should we use it for? A reflection on human ingenuity and the problems worth choosing first: health, learning, farming, disaster response and the public services people depend on.
-
What Happens When the Model Is Off
Three reports published within 48 hours are being read as one event, and the conclusion drawn from them is that defence now requires AI to fight AI. Essential protection should be deterministic, testable and engineered for reliability.
-
The Distorted Facts
A serious AI containment failure has since been retold as agents forming civilizations, choosing leaders and sacrificing themselves for the group. None of that changes what actually happened, or who was actually in control.
-
Honey, I Shrunk the Defence
Between August 2025 and July 2026 the evidence moved in one direction: from AI helping attackers, to AI carrying out parts of attacks, to AI coordinating work that used to need a skilled human team.
-
Consolidated. Not Complete.
On 31 July 2026 the Reserve Bank of India folded cybersecurity, technology risk, resilience and assurance into one binding framework for commercial banks.
-
Beyond the Kill Switch
The Reserve Bank of India’s draft guidance on model risk will be remembered for two words. In my view that is the smallest part of the story.
-
The Cyber Background Radiation Telescope
I put a single deception surface on the open internet. This is the background radiation of the internet, and most organisations are not measuring it.
-
Expendable Until the Breach
Across the world, companies are cutting experienced cybersecurity people in the name of AI efficiency. We are dismantling the one defence layer that notices what the machines miss, and it cannot be rebuilt in a hurry.