Sir Oliver Dowden

My right hon. Friend is of course entirely correct. Members on the Labour Back Benches are, given their presence in the Chamber for this debate, at the more hawkish and national security end of the left-wing spectrum, but there are elements of the liberal left who succumb to that equivalence argument and it is profoundly dangerous. The 2015 review of the Muslim Brotherhood concluded that membership of, association with, or influence by the Muslim Brotherhood should be considered a possible indicator of extremism. A decade later, in the current far more dangerous geopolitical environment, we need to look again at that assessment. That is why I continue to believe that the Government should take up the opportunity to do so. Some of our closest allies have done so and indeed have taken a much more robust approach. For example, the United Arab Emirates has designated the Muslim Brotherhood as a terrorist organisation. We should ask ourselves why a country such as the UAE has such concerns about the radicalisation of its own students in our mosques that they are restricting their study here. The test under this Bill is rightly a foreign powers threat test, and there may well be other avenues for addressing this issue, but none the less I make the point that where an ideological body or a body linked to an ideological movement is acting to the benefit of a hostile state, this regime, alongside others, should apply. Finally, I would like to address the future-proofing of this legislation, which has been touched on by some Members. Just as threat actors have evolved, so too have their methods, and foreign powers and their proxies no longer exclusively rely on the old-school methods of human surveillance, forged documents and so on. They can deploy bots that distort the democratic debate, as we have seen; ransomware to cripple business; cyber-attacks to disrupt public services; and deepfakes to impersonate trusted public figures. That list goes on, and AI will accelerate all of that—a point I made at the UN General Assembly ahead of the Bletchley Park AI safety summit. Cross-Government co-ordination to address this issue is more urgent than ever. In this country we cannot just look at state threats purely in a physical silo, detached from the digital domain. Hostile states will use AI to automate reconnaissance, scale spear phishing, identify vulnerabilities, and so on. A criminal who once needed specialist skills may increasingly be able to purchase or access those capabilities through AI-enabled tools, and hostile states that once needed sophisticated cyber-units may be able to outsource, automate or accelerate parts of that work. Many Members will be familiar with the concerns and indeed opportunities surrounding Claude Mythos and its advanced cyber-capable AI models. That is a warning to us, but it is a warning of much more to come in this space.