Government finally signals action on AI

This article is an on-site version of our Inside Politics newsletter. Subscribers can sign up to get the newsletter delivered every weekday. If you’re not a subscriber, you can still receive the newsletter

Good morning. I’m Lucy Fisher, the FT’s Whitehall editor and host of the Political Fix podcast, covering for Stephen while he’s away.

PMQs is back today after a short hiatus.

I’m fascinated to see whether Kemi Badenoch goes after Andy Burnham over his loose talk, which is turning out to be something of a (bad) habit.

His remarks at the weekend about Russian interference in the 2016 EU referendum prompted criticism that he was framing the Brexit vote as illegitimate, a suggestion Downing Street was forced to knock down.

Other recent comments by Burnham have also raised eyebrows, including among Labour officials. These span remarks that appeared to introduce the prospect of a Scottish independence referendum and to cast doubt over a Premier League investigation into Manchester City.

Badenoch quipped on X that the “busiest job in Westminster is running after the PM with a dustpan and brush, sweeping up whatever he said five minutes earlier”. Let’s see if she warms to that theme in the Commons showdown at noon.

Inside Politics is edited by Harvey Nriapia today. Read the previous edition of the newsletter here. Please send gossip, thoughts and feedback to insidepolitics@ft.com

Shifting into gear

For some weeks now, I have marvelled at just how little airtime the challenges posed by AI have received in Westminster.

The Hugging Face attack that saw rogue OpenAI agents escape their guardrails, the tech company whistleblowers warning of potential doom within a decade, the shock move by Anthropic to withhold a powerful new model from Britain’s AI Security Institute (Aisi) for testing (a decision thankfully since reversed) . . .

All were big stories in recent months that seemed to garner curiously little attention among ministers and MPs, either in parliament after summer recess or during conference season as parties gathered to reflect on their priorities.

A long weekend at Cheltenham Literature Festival deepened my suspicions that my consternation is widely shared.

I took part in three events, all loosely focused on the UK political landscape, and was struck by the preponderance of audience questions about politicians’ grip on AI, from driverless taxis and widespread job losses to how to tax robots. People seemed exasperated that the technology’s perils and predicaments were not featuring more prominently in the national debate.

The loudest cheers resulted from a rallying cry by Labour MP Jess Phillips — one of the Commons’ best communicators — exhorting ministers to simply get on with regulating tech companies.

This week, however, it feels as if both the government and parliament have finally turned a page on the topic.

On Monday, AI minister Kanishka Narayan declared that the technology must face regulation, as he set out plans for compulsory safety reporting and monitoring of frontier AI, as well as a potential requirement to make models available for pre-release monitoring.

He warned that a breach of Britain’s cyber defences by rogue agents was “likely, if not inevitable”, and urged frontier AI labs to avoid releasing models if they cannot “adequately control the risks”.

Yesterday, a second significant intervention came via MI5’s director-general Ken McCallum, who dedicated a significant proportion of his annual threat update to the dangers of AI and the need for more technological resilience. AI is “already supercharging the cyber security contest between attackers and defenders . . . I’m not going to pretend this is a relaxing contest”, he said.

He detailed the technology’s ability to “produce propaganda and radicalising material tailored to the individual”, improve terrorists’ reconnaissance and attack methodologies and enhance or create new bioweapons. In response, he announced that MI5 has launched the AI Threat Assessment Centre (Aitac) to bring together expertise from the Aisi, the National Crime Agency and the National Cyber Security Centre (an arm of GCHQ).

But he was unstinting in his message that the world is entering an uncharted era: “Beyond misuse of AI by human adversaries sit the potential risks posed by AI itself . . . Human history has given us no experience of facing an actor which is smarter than we are. Is capable of out-thinking us. This is genuinely novel.”

It seems McCallum and fellow intelligence chiefs issued a wake-up call to the cabinet last month too. During a briefing to the prime minister and senior government colleagues, the MI5 head explicitly warned that AI was “the biggest thing you should be thinking about”, according to The Sunday Times.

An increased sense of urgency has meanwhile emerged in parliament, with the business and trade select committee taking the lead on scrutiny. Yesterday, it held a marathon three-hour session on AI adoption, safety and governance, hearing evidence from a number of officials and industry figures. Liam Byrne, the Labour chair of the committee, said the oral hearing marked the start of a 12-month inquiry into the issue, which will produce regular reports.

A couple of eye-popping news stories that emerged this week have further underscored why a quicker pace is needed.

Anthropic warned that rogue “AI worms” will soon be able to hijack smart home devices such as ovens and televisions, spreading slowly across insecure appliances like “digital slime mould”(!).

As if on cue to demonstrate the kind of havoc that a network of rogue autonomous worms could wreak, a cyber attack hit French supermarket fridges and raised their temperatures, melting equipment and ruining produce. This has, unsurprisingly, sparked alarm about food security.

So it is welcome to see the UK government, intelligence agencies and parliament visibly shift into gear to take bolder action, and a reminder that public sentiment can often run ahead of political debate.

Now try this

I’m excited for the premiere next week of Laurence Osborn’s new music — and not just because (disclosure alert) he’s a friend. A highly imaginative and original composer, his latest work Sun Sundered Head: Nocturne for Cello and Orchestra will be performed by the Royal Philharmonic at Cadogan Hall next Wednesday.

If you’re intrigued, but not London-based, you can check out his latest album Tomb! on Spotify. This is how his label pitches it: “A harpsichord leading a string quartet in a juddering dance; Orpheus seeking passage into the underworld; a fantastical parade of images caricaturing musical Romanticism.” It’s gripping, evocative and like nothing else I’ve heard before.

Top stories today

• Road to referendum | Questions about the path to independence from the UK will be thrust into the spotlight at the Scottish National Party’s annual conference this week.
• ‘Scorecard positive’ | John Healey is expected to pay for new giveaways in the Budget, including a cost-of-living package, with higher taxes or spending cuts rather than more borrowing.
• Means business | Andy Burnham has told leading British bosses he wants to run the most “pro-business country in the world” in an attempt to calm nerves about tax raids.

Road to referendum | Questions about the path to independence from the UK will be thrust into the spotlight at the Scottish National Party’s annual conference this week.

‘Scorecard positive’ | John Healey is expected to pay for new giveaways in the Budget, including a cost-of-living package, with higher taxes or spending cuts rather than more borrowing.

Means business | Andy Burnham has told leading British bosses he wants to run the most “pro-business country in the world” in an attempt to calm nerves about tax raids.

Recommended newsletters for you

The Week Ahead — Start every week with a preview of what’s on the agenda. Sign up

Newswrap — Our business and economics round-up. Sign up