National Security (State Threats) Bill

Lord Garnier Excerpts

Committee stage
Tuesday 30th June 2026 Lords Chamber

There are no unintended consequences here. I know that the Minister has been listening to the arguments, because I met him again today and he showed enormous courtesy and patience in dealing with this. I am only sorry that we have not had longer in Committee, where we could have debated these points in granular detail, rather than at this breakneck pell-mell speed. There is an old saying that if you legislate in haste, you end up repenting at leisure. Let us hope that repentance will not be required on this occasion.

[Lord Garnier (Con)]

My Lords, I thank the noble Lord, Lord Alton, and the noble Baroness, Lady Hayter, for what they said. They both spoke in very measured terms and I wish to support broadly everything that they have said, for two reasons: first, because I agree with them, and, secondly, because I was too late to co-sign their amendments. The only point of difference I have with the noble Lord, Lord Alton, is some thoughts about SLAPPs legislation—I confess that, yes, I am a member of the defamation Bar and I have various views about the questions that lie behind the excitement around the need to introduce SLAPPs legislation, but that is not for today. I broadly agree with the noble Lord and the noble Baroness and give them the encouragement that, simply because a Conservative Member of this House has not signed their amendments, that does not mean that a Conservative does not agree with them.

There is a danger that we can become naive, thinking that because there are, say, 2 billion buyers of toothbrushes in the People’s Republic of China it would be wrong of us to criticise the Chinese Government when they do things, either through their own agencies or through their proxy agencies, to damage the well-being of this country. As the noble Lord, Lord Alton, calmly pointed out, not everything is always as it seems. Therefore, although neither the noble Baroness, Lady Hayter, nor the noble Lord, Lord Alton, has threatened a vote on their amendments, I urge my noble friend the Minister—and he is my friend—to bear carefully in mind that when we are dealing with China we are dealing with an underwater crocodile. We can be naive and keep feeding it, and allow it to build bigger and bigger embassies and to spy, either formally or informally, upon the activities of our citizens, and indeed of the expatriate citizens of Hong Kong in this country, and we can allow it to carry on behaving appallingly towards United Kingdom citizens, such as Jimmy Lai, who “live” in Hong Kong, but, if we let another week or month or year pass in which we live in this fog of confusion, we are, I suspect, misleading ourselves and not doing our own citizens or the citizens of other countries any good.

China is not anything other than something that we need to be very careful about. Yes, we can sell 2 billion toothbrushes and we can sell glass, IT or whatever it may be, but I urge the Government to please not allow themselves to think that, when you feed a crocodile, it will wag its tail and say, “Thanks very much”. It will not. It will say, “Thanks for your leg—I’ll have the other leg, and then I’ll have your arm, your other arm and your head”. Meanwhile, people such as Jimmy Lai, who stand up for freedom of expression, are left incarcerated in terrible conditions in Hong Kong.

Let us wake up and realise that we are dealing with adversaries, not with people who wish to do us a favour. I thank the noble Baroness, Lady Hayter, and the noble Lord, Lord Alton, and the noble Baroness, Lady Doocey, who has yet to speak, for their amendments, because they are seriously worth considering.

[Baroness Doocey (LD)]

My Lords, I will speak to Amendment 5, in my name and that of my noble friend Lord Marks of Henley-on-Thames, and the noble Lord, Lord Alton of Liverpool. This amendment seeks to ensure that our national security framework is not only addressing the threats of today but is prepared for what the director-general of MI5 has described as “the next frontier”—the potential risks posed by non-human autonomous AI systems. While our current debate focuses largely on human adversaries and their proxies who seek to undermine our national security, we must now also grapple with the emergence of systems capable of evading human oversight and control.

We are drawing a distinction between the AI tools we use today and what experts term artificial superintelligence, or ASI. The amendment defines ASI as a system

“capable of adapting its behaviour, in response to … its … environment, in ways not fully specified in advance by its developers”.

As Lord Browne of Ladyton, a former Secretary of State and expert on non-proliferation, has explained previously, the key distinction is that ASI would be an agent rather than an instrument. While a nuclear weapon cannot improve or replicate itself, or resist being shut down, a superintelligent AI system could, in principle, do all three. We are therefore contemplating a technology that could out-think experts and potentially subvert even our most capable national security apparatus.

This is not science fiction. We have recently seen a leading AI company restrict its latest, most capable model to a small group of major technology infrastructure firms, precisely so that they could use it to identify and close security loopholes that it was able to uncover in the core software and system. Officials are now warning that, in the wrong hands, the same kind of system could dramatically speed up cyber attacks against individual companies and, potentially, the critical infrastructure on which the country depends. That is where we are already today. The next level may be artificial superintelligence, where these systems become much more powerful and the risks in the hands of hostile states or state-sponsored actors are correspondingly greater.

The implications of that decision go further still. The United States Government have now imposed export controls on Mythos, prohibiting its use outside American jurisdiction. That is a remarkable step. A foreign Government have, in effect, decided that a single AI system is too powerful, too consequential and too potentially dangerous to allow unrestricted international deployment. That tells us that at least one major power has concluded that frontier AI capability is now a matter of national security, not only for the country that hosts it but for every country that might seek access to it. If the United States Government are drawing that line, it would be quite extraordinary if we in Parliament are not asking what our own response should be.

The AI Security Institute has warned that our current trajectory risks an irrevocable loss of control over systems that are becoming deeply embedded in our national infrastructure, economy and defence. Yet the AI Security Institute lacks the statutory authority to require disclosure, compel compliance or impose any binding obligations on the companies developing these systems. We have a body that can see the risks but cannot act upon them.

More importantly, the Government’s own AI Scenarios 2030 report, published earlier this month, acknowledges that highly capable AI could give rise to

“serious, potentially even existential harms”,

and, in the absence of effective intervention, highlights autonomous risks in which advanced systems may develop objectives that “resist human control”. My amendment would bring the development of such systems by a foreign power within the definition of “foreign power threat activity” under the National Security Act 2023. It is directed at systems capable of neutralising or subverting our relevant human authorities, which I have defined as our Armed Forces, intelligence services, police and the Government themselves.