-
A model’s refusal is a warning light, not a brake. Why refusal cannot carry assurance, and what must stand in… Read more
-
Receipts, Please, Part II. The most dangerous agent may not be the most capable one. It may be the one… Read more
-
Receipts, Please, Part III. The laboratories selling cyber-defence also ship the capability that makes the defence necessary. Read more
-
Receipts, Please, Part V. Every high-value instruction should carry its own proof, at runtime, or it should not execute. Read more
-
Receipts, Please, Part IV. Detection loses. Process wins. Ferrari stopped a deepfake with a question about a book; Singapore has… Read more
-
We can build agents we cannot assure. A field guide to governing, securing and assuring AI agents: the fraud, the… Read more
-
Why the AI labs will secure your code, and still leave your company exposed. The next great fraud will not… Read more
-
Inside the Safety Stack, Part 2. Boards are buying safe models the way homeowners once bought fireproof paint. Most model-level… Read more
-
AI agents are the most powerful incentive-optimisers ever built. Mechanism design, or “reverse game theory”, proposes making honesty their rational… Read more
-
Cognitive surrender is not a skills gap that training will close; it is a governance gap that only architecture can… Read more
