An essay by Burak Oktenli argues that consequential government AI should have a named human decision owner, a meaningful route for people to challenge outcomes, and expiry controls for exceptional permissions. The practical point: automation can speed public services, but it should not make responsibility harder to find.
Watch Desk analysis
What happened
Oktenli notes that 35 of 36 countries surveyed for the OECD’s Digital Government Outlook 2026 use AI in at least one area of government, while only eight report citizen-feedback or complaint mechanisms relating to government AI. He proposes three safeguards: identify an official who can investigate and correct consequential decisions; make appeals capable of changing an outcome; and ensure temporary emergency permissions actually expire.
That expiry safeguard is for exceptional powers, not a call to put ordinary government software on an arbitrary shutdown timer. Oktenli suggests embedding requirements in procurement contracts and testing them through audits. His examples include Canada’s guidance on agentic AI and Britain’s Algorithmic Transparency Recording Standard.
Why it matters
When several automated steps feed into a human decision, a neat-looking file can conceal where an error began. A named official with access to the relevant records gives a person somewhere to take a disputed decision. An appeal that simply sends the case back through the same automated process is, as remedies go, a rather elaborate way of going nowhere.
The essay also distinguishes a temporary emergency permission from routine use of an AI system. If extra access is granted for a crisis, the permission should have a defined end and stop working when that end arrives. A public statement that a programme has ended is not, by itself, proof that its software permissions have expired.
Our read
This is an opinion piece, not a new government rule or a claim that every public-sector AI system works the same way. Its strongest contribution is turning “accountability” into practical questions: who can correct a case, how can someone challenge a decision, and what prevents temporary authority from quietly becoming permanent?
Those questions belong in procurement and review processes before a public service goes live, not in a maze a frustrated applicant has to solve afterwards. Read Oktenli’s essay.
What to watch
- Whether public agencies put named decision owners and workable appeal routes into AI procurement.
- Whether emergency permissions have technical expiry controls, not just stated end dates.
- Whether audits test how quickly a disputed decision can be traced and corrected.
Discussion spark: Should a public agency be allowed to use AI in a consequential decision unless it can name an official who can explain and correct the result?
Sources and evidence
- Source update (5 November 2026, 21:10 UTC)
- Source update (5 November 2026, 21:10 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.