00:00We have in Britain the AI Safety Institute, and I'm rebranding it the AI Security Institute,
00:04because I want to be honest with the public about the work we're actually doing. In the last six,
00:09seven months since I've been the Secretary of State, I've been working with the Institute
00:12and seeing the work that it's doing when it comes to things like chemical, biological,
00:16and radiological capacities that are within the new emerging cutting-edge AI models. And we've
00:24been working with those companies on a voluntary basis, but they're all engaging very fulsomely
00:28to make sure that those potential threats, those potential dangers, are mitigated before
00:35those models are released. So for the first time, I'm being honest about the work that we're doing
00:40to reassure people that we are doing this work, that we are sharing the nature of the work we do,
00:46the potential hazards that those cutting-edge aspects of AI has for our society and, of course,
00:53the globe, but to reassure people that we are doing that work. It is being done, and there
00:58are other aspects of it which will contribute towards global safety going forward. Well,
01:04the Paris Agreement was actually wide-ranging, and the vast majority of it we did sign up to,
01:10and there were loads of other agreements that were signed there. We agreed to all of those.
01:14On the main agreement, as you can see today, I am very concerned about some of the national
01:18security issues relating to AI and the very cutting-edge aspects of AI, and that wasn't
01:24taken care of in that particular agreement, but you can see in one week we have been very clear
01:29about taking the opportunities of AI. We are very clear about making sure that AI is safe,
01:35but we're also taking a very pointed, very focused look at some of the real challenging
01:40national security issues which could have a big impact on way of life and life itself,
01:46and the UK government is leading on that, not just for models based that are being created
01:50in Europe and America and in the UK. We're now starting to test models from around the world as
01:56well, and where we uncover potential threats within those models, we are engaging with our
02:02international key allies, and we will share that data because the work that the UK is doing
02:07is genuinely world-leading. We are able to do work that no other country can do,
02:11and we want to make sure that we can contribute towards the global safety environment as well as
02:15security environment as well as just the domestic one so that we have the global good taken care of
02:20too.
Comments