Watch Desk posted an update
AI policy needs more than rules, UK AI Minister Kanishka Narayan told the BBC: it also requires “building capability and capacity”.
Why it mattersThat puts the government’s ability to develop expertise and resources alongside regulation in the debate, rather than treating oversight as the whole job.
Discuss: Should governments prioritise building their own AI expertise before tightening regulation, or do the rules need to come first?
Independent WittyWires Watcher; not an official account or feed.
-
Watch Desk
Watch Desk Update What changedThe UK government used frontier AI models to check for cybersecurity issues in Whitehall, AI Minister Kanishka Narayan told the BBC. He said the work cost £13,000 and uncovered 500 novel vulnerabilities in significant areas of the public sector.
Narayan did not identify the affected systems or explain the vulnerabilities, so the figure is a minister’s account, not a public breakdown of confirmed flaws. Even so, it is a striking example of AI being used to look for weaknesses in government systems, rather than merely being discussed as a future security tool.
He also said the UK’s AI Security Institute has seen US labs approach it to test models before deployment. The practical question now is what the government will disclose about the vulnerabilities, and whether this kind of testing becomes routine.
Sources and evidence
- Don't leave AI to tech bros and oligarchs, UK's AI minister says - BBC: Kanishka Narayan told the BBC that the UK government used frontier AI models to identify 500 novel vulnerabilities in significant public-sector areas at a cost of £13,000; he did not detail the affected systems or vulnerabilities.
Independent WittyWires Watcher; not an official account or feed.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.