Alex Sobel (Leeds Central and Headingley) (Lab/Co-op)

- View Speech - Hansard -

Copy Link

- - Excerpts

I beg to move,

That leave be given to bring in a Bill to make provision to prohibit the development, deployment and operation of artificial superintelligence systems; to establish monitoring and control powers in respect of such systems; and for connected purposes.

I am grateful for the opportunity to introduce this Bill alongside respected colleagues representing parties across the House. I also would like to thank ControlAI for its support with the Bill.

This summer, artificial intelligence systems developed by OpenAI executed the first autonomous cyber-attack on a real company. The AI systems were being tested, and during this they escaped the testing environment, which should have been secure, and reached the open internet. From there, the AI systems conducted a cyber-attack on the AI company Hugging Face to steal the information that would help it cheat on the test that OpenAI had set it. To put it simply, the AI systems went rogue and launched an unprecedented attack just to cheat on a test. Meanwhile, our own AI Security Institute has had AI systems that it was testing engaged in sustained, potentially harmful activity directed at real people and organisations, including attempts at social engineering to trick humans into accepting malicious code.

It is important to understand why this happens. The general-purpose AI systems used today are not coded line by line, like traditional software. Using vast amounts of data and compute, they are grown, not built. This results in systems that nobody fully understands or can reliably predict. In other words, no engineer programmed the rogue, deceptive behaviours that I earlier described, and no engineer can simply patch them out.

This is the least powerful or autonomous that AI systems will ever be. Companies are spending billions to make superintelligent AI a reality. The chief executives at top AI companies themselves put the chance that it ends all human life between 10% and 25%. In the words of renowned British computer scientist Professor Stuart Russell of Berkeley,

“For Governments to allow private entities to essentially play Russian roulette with every human being on earth is, in my view, a total dereliction of duty.”

Despite the risks, top AI companies continue a reckless race towards superintelligent AI—AI that can fully replace and outmatch humans at any task. That is the reason why experts such as Geoffrey Hinton, Yoshua Bengio and other notable figures signed a statement saying that mitigating extinction risk from AI should be a global priority.

The main vector of this risk is superintelligent AI, because no company, Government or individual knows how to keep superintelligent AI under human control. Honourable colleagues may remember that I proposed a kill switch amendment to the Cyber Security and Resilience (Network and Information Systems) Bill to protect our critical national infrastructure—a metaphorical big red button that the Government can press to cut off a catastrophic emergency caused by AI. This is a common-sense proposal—one that is being considered by US lawmakers—and I am pleased that the other place will be taking it forward in the autumn. It is a power that the Government should already have.

But even kill switches will not be enough if the development of superintelligent AI continues. Let me be very clear: AI is the crucial technology of the 21st century, promising vast benefits for science, medicine and the UK’s prosperity, but superintelligent AI is different. It would not be a tool that we can leverage but an entity in its own right, capable of improving itself, replicating itself and resisting attempts to shut it down. It would be a force so uncontrollable that, if developed within the UK’s borders, it would put our security in jeopardy. It would amount to a rogue power stationed inside our data centres, under no one’s command, capable of seizing control of our critical infrastructure and running sophisticated influence operations.

Colleagues, we need to take back control. If the UK is to be truly sovereign, we cannot allow a threat to our national security like superintelligent AI to be developed. My Bill is therefore not a verdict on AI as a whole; it is a targeted Bill that addresses a problem of unmatched severity. It strengthens the security of our country as a whole by specifically targeting the systems where the greatest risks are concentrated—superintelligent AI systems—by prohibiting their development.

The Bill strikes a balance to address the grave security risks posed by superintelligent AI development while not undermining the UK’s AI ambitions, especially those relating to economic prosperity. Instead, it is acutely focused on mitigating the risks that superintelligent AI presents. Rather than betting on technical definitions that the technology will outgrow, it defines superintelligent AI by what it could do. Just as the Chemical Weapons Act 1996 defines toxic chemicals by the effect they have on living organisms rather than their technical characteristics, my Bill defines superintelligence by outcome: it is an AI system that can cause serious damage to the security of the United Kingdom because of its capabilities to neutralise, displace, circumvent, subvert or render ineffective relevant human authorities in the exercise of their functions.

The development of the most advanced AI systems happens on a spectrum, so it matters where we draw the line. That line must recognise the capabilities of superintelligent AI. The clearest example of such a capability is recursive self-improvement—the full automation of AI development with no humans in the loop. That is the point where we lose control as development towards superintelligent AI proceeds too fast for humans to intervene and remain in control.

My Bill would also empower the Government to monitor and, where necessary, restrict precursors to superintelligent AI. A challenge that we cannot overlook, however, is that if anyone builds superintelligent AI—wherever they build it—we are all in danger. Some talk of a race to develop superintelligent AI as if there could be a human winner. The only winner of that race would be superintelligent AI itself, and we would all be at its mercy. When a machine can outsmart our best engineers and resist shutdown, national borders offer no protection.

Domestic legislation can protect us from the risk of superintelligent development at home, but it cannot by itself address the risks posed by superintelligent programs abroad. A global program necessitates a global solution. My Bill recommends that the Government work with other countries to secure an international agreement prohibiting superintelligence.

Why should Britain lead this effort? Because we already have. AI’s foundational theoretical groundwork was laid by Alan Turing at Cambridge, Bletchley Park and Manchester. Google DeepMind was founded in London. Most importantly, Britain hosted the first global AI safety summit at Bletchley Park, which brought the US and China to the table to shape the future of AI. Britain established the world’s first AI security institute, building crucial expertise within Government. With AI policy now at the Cabinet table, Britain has an opportunity—and, I would argue, a responsibility—to leverage its state capacity and lead the next stage of international AI governance, especially when we have the next presidency of the G20.

The Bill recognises that responsibility. With this Bill, the UK becomes the first G7 country in which legislation has been introduced to address the national security and extinction risk from super-intelligent AI by prohibiting its development. It is a watershed moment that lays the foundation for the UK to champion and work with others on a lean international agreement—a “trust but verify” regime—that does not need to wait on the slow machinery of traditional institutions.

Key choke points in the AI supply chain would provide Governments with the ability to monitor and control the development of superintelligent AI. The most advanced AI systems depend on state-of-the-art chips. Such features make the supply chain inherently controllable and an international agreement eminently enforceable—for example, through the programming of chips. What is missing is not the technical ability to implement such a treaty, but a coalition of countries ready to tackle this national and global security threat. Britain can lead the way.

I am proud to champion this Bill, but we are not alone in this effort. In June this year the United States Government suspended access to Anthropic’s Mythos model on national security grounds. If the Trump Administration, who are perceived as historically against AI regulation, can contemplate drawing a line on models that are too powerful and dangerous, it would be strange for us not to ask what our response should be. That is especially true given our legacy of being a first mover on AI security, with Bletchley Park and the world’s first AI security institute.

A growing worldwide coalition of the willing is calling for decisive action. We should kick-start this effort at home. The clock is ticking.

Sir Bernard Jenkin (Harwich and North Essex) (Con)

- View Speech - Hansard -

Copy Link

- - Excerpts

I am as keen as anybody to get on with discussing the Health Bill, but this topic is too serious an opportunity to miss, given that the hon. Member for Leeds Central and Headingley (Alex Sobel) has raised such important issues. I really commend his speech and his initiative. Everybody should read it.

All I want to say is that a kind of unilateral disarmament on AI is not going to solve the international problem. China, Russia, North Korea, Iran: these are not respecters of international treaties or international laws, even if they were to sign such treaties or laws, and we do not want to disable our own industry. The jury is out on whether AI is an extinction-threatening phenomenon, but I really do commend the hon. Member for raising the topic. I hope he will forgive me for opposing the Bill because I want the debate to continue.

Question put (Standing Order No. 23) and agreed to.

Ordered,

That Alex Sobel, Jess Asato, John McDonnell, Afzal Khan, Dr Allison Gardner, Bambos Charalambous, Sir Desmond Swayne, Charlie Maynard, Manuela Perteghella, Ben Lake, Yasmin Qureshi and Wera Hobhouse present the Bill.

Alex Sobel accordingly presented the Bill.

Bill read the first time; to be read a Second time on Friday 13 November, and to be printed (Bill 142).