Written by Rahim Hirji, author of SuperSkills (Kogan Page, 2026), founder of The SuperSkills Intelligence Company. The research is published openly with every source graded. Found an error? Tell me and it is corrected on the page.
Home / Keynotes / AI agents and accountability
For boards, operations and technology leadershipEvery oversight model ever designed assumes a pause between the recommendation and the act. Agents remove the pause. That is a different problem from the one most AI governance was built for.
Watch the showreel · 2 minutes
A feel for the room before you put me in front of yours.
The distinction is not about capability, it is about sequence. A system that answers gives you something to reject. A system that acts has already done it, and your options are limited to detection and repair. The argument is at should I let an AI agent act on my behalf.
Singapore's regulator is unusually honest about the limit. Its agentic AI framework of January 2026 concedes that continuous human oversight over all agent workflows becomes impractical at scale, which is the concession most governance documents avoid making.
Which decisions may be delegated, to what, under what conditions, and who is answerable when it goes wrong. That is a work-design decision rather than a technology decision, and it is usually made by accumulation rather than by anyone in particular. The working artefact is at the delegation boundary map.
The accountability question is genuinely unsettled and the keynote says so. What can be said is set out at AI agents and human judgement: the four roles that survive when agents act, and the named person who has to be able to explain the result.
The sequence. A system that answers gives a person something to reject before anything happens. A system that acts has already acted, so oversight becomes detection and repair rather than approval. Most existing oversight models assume a pause that agents remove, which is why they transfer badly.
Genuinely unsettled, and this research says so rather than offering false confidence. What can be stated: DIFC Regulation 10 in the UAE reasons that where an autonomous system operates for its deployer, its position is substantially similar to that of an employee, and makes the deployer liable accordingly. That is a liability analogy rather than a competence requirement, and it does not settle who inside the organisation must be able to explain the decision.
No. It is about delegation, accountability and what happens to a team when agents do the coordination. For a technical briefing on how agents are built, this is the wrong speaker and the guide to choosing an AI keynote speaker will point you at a better fit.
Written by Rahim Hirji, author of SuperSkills (Kogan Page, 2026), founder of The SuperSkills Intelligence Company. The research is published openly with every source graded. Found an error? Tell me and it is corrected on the page.
Tell me the audience, the date and the decision. A reply within 24 hours.
Also on human oversight and accountability and for boards and leadership offsites. Browse every topic, audience and region.
No enquiry today, or not yet the right time? The weekly letter is the low-commitment version of the same argument. Weekly essays on AI, capability and the future of work. Read by 25,000 people, every week since 2017. Free, and one click to stop.
Opens Substack to confirm. No pitch in it, unsubscribe in one click, and nobody follows up because you read something.
