B

Baroness Lloyd of Effra (Lab)

Speaking in the House of Lords on 26 October 2026

Debate

Cyber Security and Resilience (Network and Information Systems) Bill

Contribution

I thank noble Lords for their engagement on the important issue of addressing AI cyber risks and I share the concerns raised by this House and the public. The Government take the safety of the British people incredibly seriously. AI risks are trans-national threats. The best response to a global issue is a global response. At the UN General Assembly in September, the Prime Minister set out that the UK will work towards agreeing a single set of global principles and standards to ensure that AI development is safe and will make AI a central focus of our G20 presidency next year. It is important to situate this Bill and its important yet targeted role within the Government’s broader approach. We will legislate where necessary, alongside all the other tools at our disposal. I will now set out how the NIS regime currently covers AI risk and applies to some AI digital services. The Bill takes an all-technologies, all-threats, all-hazards approach to protecting regulated services. This means that essential services need to manage cyber and physical risks associated with AI systems underpinning their services and protect themselves from threats. Many digital and managed services within scope of the regime are enabled by, deploy or help to deliver AI capabilities. This includes cloud computing services, which provide software, platforms and infrastructure services that many organisations across the economy rely on. Commercial AI services are often delivered under similar digital service models and so are likely to be in scope. This includes AI assistance delivered via software-as-a-service models, AI platforms, and infrastructure used for AI workloads. Where an AI digital service falls in scope, the provider will have to take steps to manage the security and resilience of their services and the systems that underpin them, just like any other regulated entity. I have met with the CEO of the Information Commission. We agree on the importance of guidance to provide industry with clarity on these definitions. The regime is not designed to provide a comprehensive framework for AI governance, including frontier AI capabilities which have not been released to customers, or wider safety or societal impacts. These are important issues that need addressing at the global level, as well as through the AI Taskforce and regulation at point of use in the UK. Where AI-centred digital services are in scope of the NIS regime—for example, registered as cloud computing—our national security powers of direction could be used to direct those regulated organisations, subject to meeting strict legal tests. The direction power allows government to direct regulated organisations to take specific action where a security or operational compromise, or threat to their NIS, creates a national security risk. Directions must be necessary and proportionate. Where the legal tests are met, the direction powers enable the Government to require certain actions in an AI loss of control scenario; this could include actions such as isolating systems, restricting data flows or suspending automated actions being taken by an affected AI system. AI models are often distributed across different data centres and jurisdictions. A broad power to shut down AI systems could have implications for investment, growth and our digital sovereignty. It must be very carefully thought through. If there is a loss of control scenario that compromises a regulated entity and poses a national security risk, directing the entity to take certain actions is more proportionate and, likely, more effective in containing the risk. This addresses the underlying risk raised in Amendment 65 tabled by the noble Lord, Lord Clement-Jones. I turn to Amendment 41 tabled by the noble Baroness, Lady Harding, and spoken to by my noble friend Lady Berger, the noble Viscount, Lord Colville, and the noble Lord, Lord Holmes of Richmond. Where AI digital services are regulated under the NIS regime, the providers of those services will have to report incidents that have affected the security and operation of their systems. This includes incidents where there is a loss of control of an AI system used to provide a regulated service and where that loss of control disrupts the provision of that service, compromises user systems or has other significant impacts on the functioning of the UK’s society and economy. We will bring forward a consultation setting out further detail on how the incident reporting regime will work, including how the regulated organisations should assess whether an incident is significant. This will consider how the framework applies to emerging technologies such as AI. However, I would like to reassure noble Lords that the Bill’s reporting requirements already provide the underlying framework for identifying and reporting significant incidents involving AI systems used in regulated services, as well as other cyber incidents affecting those services. Indeed, these requirements place the onus for reporting on the regulated entities affected by an incident. This is in keeping with the overall principle of the NIS regulations, which are about how the UK’s most essential services can manage risks to their systems. It is also one of the reasons why I cannot support Amendment 41 from the noble Baroness, Lady Harding, which would place the onus for reporting on developers of frontier models rather than regulated entities. Regulating the developers of frontier AI is a different question, and I do not believe that this Bill is the right legislative framework in which to answer it. The Government will, of course, continue to take action where necessary to keep the UK safe, but these risks and technologies are global. Placing duties on developers operating in the UK will not by itself keep us safe from developers and models elsewhere in the world. Turning to how we plan to strengthen the Bill and the NIS regime to better understand and protect against AI cyber risks to our essential services, the Government have tabled an amendment to the delegated powers in the Bill. Clauses 30 to 35 list examples of what provisions may be made under Clause 29 and set out certain requirements and restrictions. Amendment 70 expands this to clarify that future regulations may enable the Government to use in-house expertise, such as AISI, and gather information on why agentic incidents in our essential services happen. Our intention is for the information to then be shared with relevant stakeholders to help prevent similar incidents. Detail on how this function could work will be consulted on. I will also bring forward three further measures in secondary legislation to manage the risks posed by agentic AI. These will be consulted on. To the question raised by the noble Viscount, Lord Camrose, I have seen the questions raised by techUK and undertake to continue to consult with industry on these matters. The measures are: first, to include technical and organisational requirements within the proposed security and resilience requirements to reduce the risk of agentic incidents affecting regulated services; secondly, to preserve information that regulated entities hold on incidents that stem from agents malfunctioning, so that others can use that information to understand why incidents happen; and, thirdly, to require regulated entities to tell other organisations—for example, suppliers—if agents used by the regulated entity are negatively impacting those organisations’ operations when an incident is occurring. Finally, the changes I mentioned today on developing a voluntary framework for regulated entities on near-miss reporting will also be a meaningful step to developing our understanding of why and how this type of incident takes place. The commitments I am making today are an example of us responding with speed and precision to address potential harms to the UK posed by novel AI risks. I strongly consider this to be the right approach. I understand that there are diverging views on how fundamental AI regulation should take place, and there is a debate to be had on this, but the Bill is not the place for it. The amendments tabled by the noble Lords, Lord Tarassenko and Lord Clement-Jones, and the noble Baroness, Lady Kidron, would provide the AI Security Institute with a statutory role to regulate the deployment of frontier AI models in the UK. These amendments would not be effective in making our essential services safer. AI is a dual-use technology. Malicious actors could use frontier AI capabilities to attack the UK, but defenders can use the same capabilities to protect it, as my noble friend Lady Berger pointed out. Delaying essential services’ access to the latest tools would leave them less well defended, while attackers would face no such delay. These amendments could hold back adoption for up to 10 weeks in a field where new models come out every few weeks. The scope of this Bill is also not right for AISI, or the regulation of frontier AI models. In particular, a regulatory role for AISI would undermine its voluntary partnership with leading AI developers, which give it early access to the most advanced models and on which its vital work to inform government depends. This is in contrast to the EU’s regulatory stance and lack of access to frontier models. Finally, as I have said, the risks from frontier AI do not respect borders. We believe that the most effective response, as the Prime Minister committed to, is agreeing international standards. This Government remain committed to making sure that AISI is equipped to fulfil its vital role, and will keep the House updated on its work. In response to the question from the noble Lord, Lord Clement-Jones, I will not pre-empt the Government’s consideration of or response to the Joint Committee on Human Rights; the response will be published in due course. I thank noble Lords for their contributions. Through the commitments I have made today and the commitments made by the Prime Minister, the Chancellor and the Minister for AI in recent weeks—which, critically, seek to build UK capability through our sovereign AI fund and other measures—I hope I have been able to show that the Government are absolutely committed to ensuring that we will do whatever it takes to keep the British people safe.

More from Baroness Lloyd of Effra (Lab)

Other recent Hansard contributions by the same speaker.

About Hansard

Hansard is the official verbatim record of proceedings in the UK Parliament. Every word spoken in the Commons and Lords is recorded and published — this page is a single contribution from that record.