PARLIAMENTARY DEBATE
AI: Regulation - 17 April 2024 (Commons/Commons Chamber)

In our White Paper on AI regulation we set out our ambitious pro-regulation, pro-innovation framework, outlining five cross-sectoral principles to be applied by existing regulators. In February we published our response, setting out how we are supporting regulators to deliver the framework and strengthen our global AI leadership. That includes new funding and guidance for our regulators, and we have established a central risk function to support.

Yet we heard just a few months ago from the Prime Minister that the UK’s answer is not to rush to regulate. The Competition and Markets Authority has been clear about the potential harms that unregulated AI could generate from baking in biases that affect certain demographics, and general purpose models that could get out of control. Why have the Government dragged their feet on safeguards for the most advanced AI models, or is the Secretary of State simply waiting for the next Labour Government to control the new AI models?

Mr Speaker, this is absolute tosh. We have led the world when it comes to AI safety. We have set up a long-term process in the AI Safety Summit, and the next one will be in Seoul in just a few weeks. We have also set up the world’s first AI Safety Institute, which is testing both pre and post deployment. We have also been clear: we will not rush to legislate. We will grip the risks and better understand them, rather than produce out-of-date legislation as a gimmick.

Tomorrow the TUC will officially launch its Bill on AI regulation and employment rights, which recognises that transparency, observability and explainability are all key elements of a fair and just workplace. What will the Government do to ensure that AI does not lead to a weakening of workers’ rights?

We want to garnish the opportunities of AI for the British public, which include the comple- mentary aspect that it can pose for jobs, especially in teaching and medicine, by taking away some of the admin and bureaucracy. We are also very realistic that technology always changes labour market needs. In 1940, 60% of the jobs we now have did not exist. That is why we have undergone a revolution in our skills system, including the launch of the lifelong learning entitlement next year.

It is all very well the Government saying that they will take their time over this response, but the point is that the Federation of Small Businesses is saying that a regulatory framework is urgent, and Dr Rogoyski of the University of Surrey is pointing out that delay could mean the UK probably having no choice but to follow the approach of the US and Europe on AI regulation. Can the Secretary of State set out exactly what the timeframe will be for regulation?

The hon. Member is getting confused between regulation and legislation. We already have a plethora of regulation and world-leading regulators that we are supporting. We were clear in our White Paper response that we will legislate—as will every nation around the world—but we want to get that legislation right. She commented on the US’s approach. We are working hand in glove with the US, and I signed the world’s first memorandum of understanding on AI institutes just a few weeks ago.

What assessments have the Government made of the United Nation’s plans to internationally regulate artificial intelligence? What are the implications for UK sovereign security?

The UK Government are committed to unlocking the opportunities of AI, while mitigating the risks. That requires both domestic and international action. The UK is a leading voice internationally, having hosted the AI Safety Summit, which delivered the world-first Bletchley declaration, as well as actively participating at the UN. That includes our proactive role shaping UNESCO recommendations on AI ethics.

I call the Chair of the Select Committee.

The Secretary of State knows that leading AI developers are expected imminently to release new, more sophisticated AI models. Can she confirm that our AI Safety Institute has had access to those models, as was agreed at Bletchley Park? Is it the case that the developers have made changes to their models where they have been requested by the institute?

I know that my right hon. Friend shares my passion and enthusiasm for this topic, as well as a desire to make sure we grip the risk. Our institute is the first in the world to be doing pre and post-deployment testing, in line with the agreement we made at Bletchley Park. I cannot get into the specifics of which models we are testing, as I am sure he will understand, as that is highly commercially sensitive information, but I can assure him and the House that where risks are found, we expect relevant action to be taken. The responsibility of developers is to ensure that their models are safe, but the Government are committed to holding them to account.

Does my right hon. Friend agree that spreading best practice in this field is perhaps the most important thing? For example, the health benefits of AI have already been mentioned, such as in the diagnosis of bowel cancer, and that is about promoting the health of the public at large. Those things need to be pushed forward with urgency. It is not enough just to try to slow things down and over-regulate.

I absolutely agree. AI has the potential to be revolutionary, especially in areas such as healthcare. That is why at the summit we announced a £100 million pot to accelerate some of our existing healthcare missions. We are working hand in hand with the Department of Health and Social Care on this important topic.

I call the shadow Minister.

AI is an incredible new technology, and it can help the NHS to save lives, but there are also risks, such as the danger of deepfakes. The Government have been warned about those risks, yet time and again Ministers have dithered and delayed, and the Government’s failure to act was highlighted in the Financial Times this week. Have the Government run out of ideas, or are they just scared of their own Back Benchers?

As the hon. Member will know, we have the defending democracy taskforce, which is dedicated to this very subject and is led by the Security Minister, my right hon. Friend the Member for Tonbridge and Malling (Tom Tugendhat). We as a team are actively participating in that, and we also work with social media companies and our international counterparts. It is something that I personally put on the agenda at the summit and that I have personally discussed in forums such as the G7. The Deputy Prime Minister is also leading the way with his AI compact. There is no easy answer to this, but we are working in a conciliatory and speedy manner to ensure that we address all opportunities and answers.