Frontier Artificial Intelligence

4. What recent assessment she has made of the potential impact of frontier AI on society.

I point the hon. Gentleman to the recent report from the AI Safety Institute in December, which showed that AI’s role in social interaction, online harms, cyber-security and the workplace is increasing. We are taking the risks of AI as seriously as we take the opportunities, working closely with colleagues in the Department for Business, Innovation, Science and Trade and the Cabinet Office. We are clear as a Government that companies cannot be allowed to mark their own homework.

I welcome the response from the Secretary of State. The company OpenAI has just created a program called Astra, which, in its own words, is the most dangerous program it has ever created because it can hack other programs autonomously. That follows other AI programs unleashing themselves from their own security system and hacking other programs simply because they can—because they speak a language that only they seem to understand. AI is inevitably our future, but with the Government handing over NHS contracts and our public service records to big American AI companies, what is the Secretary of State doing to ensure that we are kept safe in this dangerous world?

The hon. Gentleman is right to point to the fact that as frontier AI develops, it poses new challenges and new risks, not just in the area of hacking but by creating confusion about legal liability, safeguards and who is responsible. There is a clear role for Government, and we are working hard to make sure that we meet those expectations.

My Select Committee’s report “Science diplomacy: Sovereignty, strategy and the global race” highlighted the risk of the UK being cut off from frontier AI models at the whim of other countries—one country in particular comes to mind. What assessment has the Secretary of State made of the impact on the UK’s society and economy if we were unable to access those models?

My hon. Friend is right to point to the UK’s need for sovereign capability in AI. I pay tribute to my right hon. Friend the Member for Leicester West (Liz Kendall), whose work I have inherited in this area. I also thank my hon. Friend for the contribution of the Science, Innovation and Technology Committee over a long period of time. We are taking the need to deal with the exposure of the UK very seriously, and I would be happy to meet her to discuss that further.

I call the Liberal Democrat spokesperson.

Frontier AI is supercharging misinformation online, but we still cannot get to grips with harms much closer to home. There are campaigners in the Public Gallery today, including Tulisa Contostavlos, who was the target of a fabricated story following an undercover tabloid sting, and is proof of the damage that an unregulated press can do. Such examples are why the Lib Dems want the full implementation of Leveson 1 and the commissioning of Leveson 2. Will the Secretary of State meet me and Tulisa to discuss finally getting proper regulation of the press?

I would be delighted to offer the hon. Lady that meeting. I take this opportunity to pay tribute to Tulisa, whom I have admired for many years for the way in which she has taken such a dreadful experience and used it to help others. Mr Speaker, can I also take this moment to acknowledge the absolutely tragic death of Professor Jason Arday, which happened since the House last met? There is an ongoing inquest and an ongoing investigation by the University of Cambridge. While the Government would not wish to undermine in any sense those ongoing inquiries, I want to reassure the House, and the many families who have been affected by the sort of practices that the hon. Lady references over many years, that we are not neutral when it comes to protecting our citizens.