TL;DR
Speaking across Labour’s conference in Liverpool, reported on Wednesday 30 September, AI minister Kanishka Narayan said he would push for legislation if its backers could show how it would deliver what voters want: safety and British capacity. He argued that building UK capability gives more leverage than rules. Several departments are nonetheless exploring AI legislation, POLITICO reports.
The minister’s challenge
Narayan framed public demand in two parts. “The thing the British public says is keep us safe and build sovereign capacity,” he said at a POLITICO event. If regulation were the route to those ends, “I would be the first one in the queue to push for that.” His question to campaigners was how their proposal actually gets there.
He was blunter about who he hears from. “I think no MPs, and almost no civil society organizations ever talk to me about how we can actually build British capability,” he told one fringe audience. At another he said the task was to fix “the error of the last 10 to 15 years of not having built capability”. He also said the UK could not afford a US-style backlash against data centre building.
A cabinet with different instincts
Labour’s manifesto promise of binding rules for frontier AI firms is still unmet, and Prime Minister Andy Burnham has said “nothing is off the table”. Three industry figures told POLITICO that multiple departments are looking at legislation; an official confirmed this but said nothing had been decided.
Colleagues pulled in different directions. Foreign Secretary Ed Miliband said guardrails could not be left to corporations. Business Secretary Jonathan Reynolds warned against measures that would “crush economic activity” and called the UK’s freedom to diverge from the EU on AI a “silver lining” of Brexit.
Not everyone is persuaded. The Ada Lovelace Institute’s Michael Birtwistle said that promising to regulate when needed while declining to commit to a bill “seems to be setting an unreachable bar for legislating”.
The same week’s other voices
Narayan is not alone in putting something else ahead of rule-writing. Bank of England governor Andrew Bailey this week argued for testing first and regulation later, but from a supervisor’s angle. Separately, technology executives have written to Narayan and Burnham asking for at least £20bn for sovereign AI, which is the capability agenda with a price tag. Earlier this week we reported Narayan’s call for a contingency plan for severe AI job losses.
Looking forward
Narayan now sits in Cabinet and says AI has come up at every meeting so far. Whether “build first” survives depends on whether departmental work on legislation becomes a bill, and on how quickly public anxiety about AI turns into pressure from backbenchers.