This month Channel 4 News filmed at UnlikelyAI office, as the UK government faced a contested AI policy decision of the year.

AI regulation in the UK: where it currently stands

AI regulation is not a single question. It is at minimum three: what uses should be restricted, what development should be permitted, and how do you enforce the distinction across borders.

The UK’s current consultation process has not yet answered any of them. William spoke to Channel 4 News on precisely that gap, not to argue against regulation, but to make the case that regulation designed without those distinctions will not achieve what it intends, and will carry costs that fall on the wrong people.

“”AI is a global race. The risk is if we have bad regulation, overly tight regulation in the UK, all it will do is drive the innovation that would have happened here to happen elsewhere.” — William Tunstall-Pedoe FREng, CEO & Founder, UnlikelyAI — Channel 4 News”

Why the regulation question is also a trust question

We have spent the last year quantifying what happens when AI is deployed without the conditions for trust in place. Our AI Trust Ceiling research found that 77% of senior leaders in regulated industries are manually verifying AI outputs before acting on them not because they distrust AI in principle, but because the systems they are using cannot demonstrate why they are right. Regulation that does not address that structural problem does not make AI safer. It just adds friction to the wrong layer.

The policy question and the trust question are the same question. AI that cannot explain its reasoning is AI that cannot be governed not by regulation, not by compliance teams, and not by the individuals accountable for the decisions it shapes. Getting regulation right means starting from that reality, not from the assumption that more rules applied to current architectures will produce trustworthy outcomes.

See the latest updates on our LinkedIn