Original Broadcaster Captioning (Enhanced). Treat it as a searchable index of what was broadcast, not a quotation record.
00:00:08MUSIC
00:00:10It's 6pm in London. This is the World
00:00:17World Today with Annita McVeigh open
00:00:19open Eye. Anthropic and Google say say they're working to create an
00:00:23an AI standards body as concerns
00:00:26concerns over the technology grow.
00:00:28grow. It comes as key US figures figures from across the political
00:00:31political spectrum attend an event event in Washington calling for kerbs
00:00:35kerbs on artificial intelligence. intelligence. China warns against
00:00:39against turning outer space into into a battlefield, as the US confirms
00:00:42confirms for the first time that
00:00:44that it's deployed space weapons.
00:00:46weapons. The EU chief hits out at at what she calls Russia's reckless
00:00:50reckless and dangerous behaviour behaviour after a drone is shot down
00:00:53down by Nato fighter jets. Train
00:00:56Train travel across large parts of of the Netherlands has ground to
00:01:00to a standstill because of what authorities
00:01:02authorities suspect are co-ordinated co-ordinated acts of sabotage. And
00:01:06And also coming up on the world today,
00:01:09today, the Emmy winning medical show show that is hitting home with real
00:01:12real life doctors.
00:01:27And welcome to the world Today, an hour of international news from
00:01:31from the BBC. Good to have you with
00:01:32with us. In the last few minutes,
00:01:35minutes, OpenAI, Anthropic and Google Google have said they are working
00:01:38working to create an AI standards
00:01:40standards body. It's reported to
00:01:42to come from a proposal by Google
00:01:45Google DeepMind's Demis Hassabis
00:01:47Hassabis for a self-regulatory body body modelled after the Financial
00:01:50Financial Industry Regulatory Authority Authority that oversees US brokers
00:01:53brokers and investment firms. President President Trump has said fears over
00:01:57over AI are a hoax, but not everyone
00:01:59everyone in Washington is so sanguine. sanguine. And key US figures from
00:02:03from across the political spectrum spectrum are attending an event in
00:02:06in Washington calling for kerbs on on the technology. This is the scene
00:02:10scene live at what is being called
00:02:13called the Pro-human Assembly. We We are expecting to hear from the
00:02:17the left wing independent Senator Senator for Vermont Bernie Sanders
00:02:19Sanders in just a moment and also also from right wing strategist and
00:02:23and Trump ally Steve Bannon. The
00:02:25The two men finding common ground ground over this issue of AI. In
00:02:29In his latest broadcast, Bannon said
00:02:32said the American people are not
00:02:34not going to be supplicants to tech
00:02:36tech oligarchs. And in some of his his latest comments on the concerns
00:02:40concerns raised by AI insiders, President
00:02:43President Trump has dismissed fears fears about AI harming humanity as
00:02:47as a scam, saying there's no need
00:02:50need for more guardrails. Well over over the last week, as you will no
00:02:53no doubt be aware, there have been been multiple warnings about AI from
00:02:57from a variety of industry insiders insiders as well as politicians.
00:03:00politicians. Let's just take a moment
00:03:01moment to run through those for you.
00:03:04you. Last Wednesday, the Anthropic Anthropic researcher Jacob Coxson
00:03:07Coxson resigned, saying the race race to build ever more powerful
00:03:10powerful AI systems could put humanity
00:03:13humanity at risk. Instead of denying
00:03:15denying these claims, Anthropic scientists scientists doubled down the company's
00:03:19company's lead for alignment science. science. In other words, the person person responsible for making sure
00:03:23sure AI works for not against us
00:03:25us said AI has a 10% chance of killing
00:03:28killing all humans by 2030 unless
00:03:31unless we gain control of superintelligence.
00:03:35superintelligence. Anthropic then then revealed it had detected attempts attempts to use its AI to potentially
00:03:39potentially aid the development of
00:03:41of dangerous biological agents. And
00:03:44And over the weekend, Anthropic chief chief executive Dario Amodei stepped
00:03:47stepped in and called for the pace pace of advanced AI development to
00:03:51to slow down. In response, China
00:03:54China criticised Amodei for saying saying Beijing's proficiency in the
00:03:57the technology could create national
00:03:59national security risks for the US
00:04:02US and yesterday Monday, President President Trump rejected those calls
00:04:05calls for controls of artificial artificial intelligence, suggesting
00:04:08suggesting they would cut the lead lead the US has over China in the
00:04:12the sector. Mr Trump said the only only guardrail that AI needed was,
00:04:16was, in his words, a strong and smart
00:04:19smart president, which the US has.
00:04:22has. OK, well, I'm just keeping an an eye on what's happening in Washington.
00:04:25Washington. Bernie Sanders not speaking speaking yet. But with me here is
00:04:28is Max Corbridge, a former ethical
00:04:30ethical hacker and now CEO and co-founder
00:04:32co-founder of AI safety company Secure Secure Genetics. Max, always good
00:04:35good to have you with us. I want want to get your thoughts, first
00:04:38first of all on that announcement announcement from OpenAI. OpenAI,
00:04:40OpenAI, Anthropic and Google that
00:04:42that they want to create an AI standards
00:04:45standards body. How confident should
00:04:47should we be in the AI industry marking
00:04:50marking its own homework? So I think think that's really the challenge challenge that we've got at the moment.
00:04:54moment. We're seeing this come out out from all different sides. We're We're seeing the government saying
00:04:58saying that the industries need to to regulate themselves. We're seeing seeing the industry saying they're
00:05:01they're doing their best. I think think what we're really seeing is
00:05:04is the start of a lot of this stuff. stuff. This is version one if you
00:05:07you think about this and really this this has been put together in essentially essentially 4 or 5 days since this
00:05:11this really kicked off. So I think think this is a good thing. I think think we need to see this from both
00:05:15both the AI providers themselves themselves and from the governments governments that can wrap regulation regulation around this. And that's
00:05:19that's really where we're going to to see the most success in the endeavour endeavour to limit the potential
00:05:23potential harm that can come from
00:05:24from AI as well. And then in the
00:05:27the last 12 hours or so, 24 hours hours or so we've heard from President
00:05:31President Trump calling all of these these concerns a hoax. Your reaction
00:05:34reaction to that? You know, some
00:05:35some people are saying this is being
00:05:39being driven by the markets. These
00:05:40These comments of his. I feel very
00:05:43very saddened by that because we we had a real golden opportunity.
00:05:46opportunity. We had a time where where we had alignment from some
00:05:49some of these frontier AI labs for for the first time. We've got the
00:05:52the idea of the CEOs of these companies companies stepping up and saying, saying, yes, we do need to be doing
00:05:55doing something about this. And this this idea that that now is the time
00:05:59time to act. And we're calling out out for things like regulation and and governments to step in. At the
00:06:03the same time, the one person who who we needed to get on board with
00:06:06with that was President Trump. And And what did he do? He came out and
00:06:08and said that it was a hoax. I think think this will be something we look look back and regret, because this
00:06:12this seems like the time where AI AI safety is on top of mind for everybody. everybody. We can and should be acting
00:06:16acting now. We've got alignment and and collaboration efforts with us
00:06:19us and China that we're looking to to work with. And so I feel that's
00:06:23that's a really missed opportunity opportunity for us. But how much much alignment can there be between
00:06:26between the US and China? Because, Because, you know, on the one hand,
00:06:30hand, you have people saying there there needs to be a framework, an
00:06:33an agreed framework between the AI
00:06:36AI superpowers. I think we're almost almost with Bernie Sanders, but we'll
00:06:38we'll try and fit in a quick question. question. No, I can see him actually
00:06:42actually arriving at the podium. podium. Max, I will come back to to you. Bernie Sanders, who announced
00:06:45announced last week a bill banning
00:06:49banning artificial super intelligence. intelligence. In other words, systems
00:06:51systems that can match or indeed indeed beat human intelligence. Let's
00:06:54Let's listen in to what he has to
00:06:56to say. He has been involved on this
00:07:00this issue before. It was fashionable, fashionable, and I want to thank
00:07:03thank him for bringing people of
00:07:06of different political perspectives perspectives together to address
00:07:10address the growing threat posed
00:07:13posed by artificial intelligence. intelligence. This afternoon, I want
00:07:17want to say a few words on what I
00:07:21I consider to be the very most important
00:07:26important aspect of this crisis that
00:07:27that we are facing, and how we can
00:07:31can best go forward. Let us be clear
00:07:37clear artificial intelligence and
00:07:41and robotics are the most transformational
00:07:44transformational technologies in in the history of humanity, and they
00:07:48they will impact the lives of every
00:07:53every man, woman and child in our
00:07:57our country AI experts have told
00:07:59told us that within a few years,
00:08:02years, AI may be able to do anything
00:08:07anything that we can do better than
00:08:10than we can do it. The economic impact
00:08:16impact of that is that AI has the
00:08:19the potential to eliminate tens of
00:08:23of millions of jobs, wiping out entire
00:08:27entire professions and making it
00:08:28it harder for our young people to
00:08:31to enter the workforce. But it's
00:08:36it's not just the economy that AI
00:08:38AI is impacting. Psychologists have
00:08:41have warned us about the increased
00:08:45increased isolation and mental health
00:08:47health challenges young people are
00:08:50are experiencing as they become more
00:08:53more and more dependent upon an AI
00:08:56AI chat bot for their emotional support.
00:09:03support. Educators are worried about
00:09:04about a decline in our kids reading
00:09:08reading scores, writing abilities
00:09:10abilities and cognitive capabilities.
00:09:13capabilities. As AI is used more
00:09:16more and more in our schools civil
00:09:20civil libertarians warn us that AI
00:09:24AI is in the process of eviscerating
00:09:27eviscerating our privacy rights by
00:09:29by making it easier to collect and
00:09:33and analyse every email that we send,
00:09:36send, every text, every website visit
00:09:40visit and every trip that we make
00:09:42make in our car resulting in the
00:09:46the creation of a mass surveillance
00:09:51surveillance state political scientists scientists warn that AI could threaten
00:09:55threaten the integrity of our elections
00:10:00elections and political institutions
00:10:02institutions go to the internet right
00:10:06right now and with sophisticated sophisticated deepfakes, you will
00:10:09will find it increasingly difficult
00:10:12difficult to determine truth from
00:10:15from fiction. And yet as disturbing
00:10:20disturbing as all of that is, there
00:10:23there is another even more alarming
00:10:27alarming danger posed by AI. And
00:10:32And that is that if AI surpasses
00:10:35surpasses human intelligence, as
00:10:37as many scientists believe could
00:10:41could happen, this technology will
00:10:44will escape human control with potentially
00:10:48potentially catastrophic consequences.
00:10:50consequences. And perhaps the most
00:10:53most important point that I want
00:10:55want to leave with you this afternoon
00:10:58afternoon is that AI is advancing
00:11:04advancing exponentially exponentially
00:11:07exponentially every single day. In
00:11:09In other words, what you are looking
00:11:12looking at right now is the least
00:11:16least capable version of AI that
00:11:20that we will ever have. The least
00:11:24least capable, the dumbest version,
00:11:26version, the speed and scope of AI
00:11:30AI development in just the last four
00:11:32four years has been almost beyond
00:11:36beyond human comprehension. When
00:11:38When ChatGPT was first released in
00:11:43in November of 2022, AI models could
00:11:45could write a short poem, a few tweets
00:11:48tweets and maybe an essay that sort
00:11:52sort of made sense. Today, AI models
00:11:56models can already escape human control
00:12:00control commit felonies by hacking
00:12:03hacking into other computers and
00:12:05and create viruses that have never
00:12:07never been seen before. And my friends,
00:12:13friends, if you are concerned about
00:12:15about what is happening today, take
00:12:18take a deep breath and try to imagine
00:12:22imagine where AI will be five years
00:12:26years from now, ten years from now,
00:12:28now, 20 years from now. And to my
00:12:33my mind, here is what is most extraordinary
00:12:37extraordinary about the transformative transformative and unprecedented
00:12:40unprecedented changes that AI is
00:12:43is making to our society. Despite
00:12:46Despite living in a so-called democracy,
00:12:49democracy, the public has had virtually
00:12:52virtually no input into the AI revolution
00:12:57revolution that is transforming the
00:12:59the world in which they live. Congress
00:13:04Congress under both Democratic and
00:13:06and Republican control, have been
00:13:08been asleep at the wheel, and we
00:13:11we now have a president whose ignorance
00:13:14ignorance regarding this issue is
00:13:16is truly embarrassing. Virtually
00:13:20Virtually all of the decisions regarding
00:13:21regarding the development of AI have
00:13:24have been made by a handful of the
00:13:28the richest people in the world Musk,
00:13:32Musk, Zuckerberg, Bezos, Thiel, Ellison,
00:13:34Ellison, Brin and a few others whose
00:13:37whose chief goal is to achieve even
00:13:40even more wealth and more power than
00:13:44than they already have. No matter
00:13:47matter what the impact their work
00:13:50work has on ordinary people. So let
00:13:53let me make a radical suggestion
00:13:56suggestion if the lives of every
00:13:59every man, woman and child are going
00:14:02going to be fundamentally changed
00:14:04changed by this technology than the
00:14:07the people of this country must make
00:14:09make the decisions about AI and not
00:14:13not just a handful of oligarchs.
00:14:25oligarchs. Bernie Sanders continuing
00:14:27continuing to speak at this event event in Washington. Max Corbridge
00:14:31Corbridge still with me in the studio,
00:14:34studio, founder of AI safety company
00:14:36company Secure Agent, a really fascinating fascinating speech from Bernie Sanders.
00:14:40Sanders. And we're expecting Steve Steve Bannon opposite end of the
00:14:43the political spectrum to speak in in a few minutes as well, a Trump
00:14:46Trump ally. But, you know, calling calling for safeguards as well. It's
00:14:50It's fascinating, isn't it, that that this subject can bring people
00:14:53people from such different parts parts of the spectrum together? Yeah. Yeah. I mean, it's one of the very
00:14:57very few topics I think, which can can and there's a chance at the end
00:15:00end of the month that we may even even bring more collaboration between between the US and China around a
00:15:04a topic like this. Just because the the risk is so great, the time is
00:15:07is now. And really it does require require everybody from all sides
00:15:11sides of the political parties as as well as both nations who are kind
00:15:14kind of really pushing this forward forward to be able to collaborate collaborate on this, to make sure
00:15:17sure we are doing the right thing. But President Trump doesn't
00:15:20doesn't think there's a risk. That's That's the issue, isn't it? He doesn't doesn't think there's a risk. He
00:15:23He thinks that if the US slows down,
00:15:25down, it will fall behind China.
00:15:28China. When there is so much to lose
00:15:32lose in terms of power and economic economic growth. As the US and China
00:15:35China see it, do you really see there there being some kind of meeting
00:15:38meeting of minds in terms of guardrails
00:15:42guardrails around this technology? technology? I think I mean, it was was interesting. Bernie Sanders there
00:15:45there calls it truly embarrassing.
00:15:48embarrassing. The element of this, this, which we are, you know, looking
00:15:51looking at now, I just think is it's it's terrifying in terms of the precedent
00:15:55precedent that we're sending with with Trump at the wheel of this sort
00:15:58sort of technology. And I think maybe maybe his initial reaction was quite
00:16:02quite explosive. I think they tend tend to be and maybe we'll see a
00:16:04a more kind of weathered approach approach and sort of metred approach approach to this as well. Very briefly,
00:16:08briefly, Max, as we have to go to to our break, Bernie Sanders posed posed the question try to imagine
00:16:12imagine where AI will be in five, five, ten, 15 years from now. You're
00:16:15You're an expert on this. Can you you imagine that? I've given up trying
00:16:18trying to think where we will be be beyond the end of 2027. And even
00:16:21even that I think I mean, if you you think about how fast we've transitioned
00:16:24transitioned and moved from the coding coding era and coding agents, how
00:16:27how quickly that's changed that industry, industry, it's coming for a lot of of other industries. And that's just
00:16:31just within the last six months. OK. Well, that answer speaks speaks volumes, I think. Max thank
00:16:35thank you very much. Max. We'll be be staying with me here in the studio. studio. We're expecting Steve Bannon
00:16:39Bannon to speak. We'll bring that that to you as well very shortly.
00:16:42shortly. But a break now around the
00:16:43the world
00:17:43interviews, results and goals as as we follow every twist of the Premier
00:17:47Premier League season here on BBC
00:17:50BBC News. China has urged the United
00:17:58United States to stop preparing for
00:17:59for war in space and respond to confirmation
00:18:02confirmation that America had deployed deployed weapons into orbit for the
00:18:06the first time. Washington said it
00:18:08it had active weapons currently circling circling the globe. The Chinese Foreign
00:18:12Foreign Ministry has warned against against an arms race in space. As
00:18:16As Jonathan Beale reports. No notice notice launch Iran. The US Space
00:18:19Space Force is already helping America
00:18:21America fight its wars on the ground ground here, detecting the launch
00:18:25launch of ballistic missiles. This
00:18:27This is the Wolf Den. But it's also
00:18:30also preparing for a potential conflict
00:18:32conflict in space. We are already already sparring in the cyber domain
00:18:35domain in the space domain. Home
00:18:37Home to thousands of active satellites
00:18:40satellites Iran. Now, for the first first time, the US has confirmed
00:18:44confirmed it's deployed a weapon weapon in space. The US Air Force
00:18:48Force secretary said it wanted to to maintain its dominance. So we've
00:18:52we've had to take steps to make sure sure that when we're threatened,
00:18:55threatened, we can take care of that.
00:18:58that. That's prompted both Russia
00:18:59Russia and China to condemn the move
00:19:02move and warn of an arms race in
00:19:05in space. America's already accused
00:19:07accused them of doing the same. The
00:19:10The reality is there's already the the start of an arms race in space.
00:19:14space. The US has accused both China
00:19:16China and Russia of developing satellites
00:19:19satellites that can jam the signals
00:19:21signals of other satellites or listen
00:19:23listen in all deploying lasers to
00:19:25to blind them. Even fitting grappling
00:19:28grappling arms to try to manoeuver manoeuver other satellites out of
00:19:31of orbit. The US has also accused accused both China and Russia are
00:19:35are developing weapons that can be
00:19:37be fired into space or from space
00:19:40space to destroy satellites. Three.
00:19:42Three. Two. We have ignition. But
00:19:45But for now, the US isn't saying saying what weapon it's launched
00:19:49launched into orbit. What the US
00:19:51US is talking about are probably probably things like electronic warfare
00:19:55warfare or jamming abilities or spoofing
00:19:58spoofing abilities, possibly things things like what we call co-orbital
00:20:01co-orbital weapons, which you have have another satellite that's orbiting
00:20:04orbiting alongside another satellite satellite and could bang into it.
00:20:06it. It's not just the military that
00:20:09that increasingly relies on space.
00:20:11space. We all do on satellites for
00:20:14for everyday life, from social media media to phone calls and finding
00:20:18finding our way about. It's not yet
00:20:21yet Star Wars, but an arms race in
00:20:24in space should concern us all. Jonathan
00:20:27Jonathan Beale, BBC News. Let's speak speak now to Juliette Java Sea, who
00:20:30who is lead on security and defence defence at the European Space Policy
00:20:34Policy Institute. Julia, welcome
00:20:36welcome to BBC News. What kind of of weapons might we be talking about
00:20:40about here in terms of what the US
00:20:43US has done? Thank you. Good afternoon. afternoon. Good evening. So regarding
00:20:46regarding the type of capability
00:20:50capability or its positioning in in orbit, we don't have that many
00:20:53many information. We know the variety
00:20:56variety and the options of counter
00:20:58counter space options that could could be developed. In this case,
00:21:02case, most likely. But just because
00:21:05because it is also consistent of
00:21:07of the efforts of the US for instance,
00:21:09instance, in relation to the banning
00:21:11banning of Dasa to kinetic weapons.
00:21:16weapons. It's most likely that it's
00:21:20it's it will be some sort of electronic
00:21:23electronic warfare or radio jamming
00:21:26jamming platforms less likely that that it's a kinetic vehicle of some
00:21:30some of some kind. And this is also
00:21:32also confirmed by really the international international commitment of the US
00:21:35US in relation to the second specific specific type of counter space. And
00:21:39And when it comes to space weapons,
00:21:42weapons, Julia, what capabilities capabilities do the US, Russia and
00:21:45and China have right now relative
00:21:46relative to one another? So Russia,
00:21:52Russia, China and the US, they have
00:21:54have all been developing manoeuvrable
00:21:58manoeuvrable satellites meant to
00:21:59to either conduct inspections or
00:22:04or potentially to then deploy payloads
00:22:08payloads which have to install sensor
00:22:11sensor payloads which have the capability capability to jam or even to direct
00:22:15direct energy weapons towards their their targets. So this is already
00:22:18already something they have, as I
00:22:21I mentioned before, they all tested
00:22:24tested also kinetic weapons, although
00:22:26although as I mentioned before, with
00:22:29with a commitment not to develop
00:22:30develop this type of to deploy this
00:22:33this type of weapons in relation relation to the debris that they
00:22:36they would be producing. And so in in some sort of a very pragmatic
00:22:40pragmatic approach also to what is
00:22:43is becoming a warfighting domain
00:22:46domain they also. Sorry, Julia, I
00:22:48I was just going to say there's a
00:22:52a 1967 treaty in place which banned
00:22:54banned weapons of mass destruction
00:22:56destruction in orbit. But what kind
00:22:59kind of guardrails? There seems to to be a word that we're returning returning to on the programme today.
00:23:02today. First, in relation to AI and
00:23:05and now space weaponry. Are there there any guardrails or treaties
00:23:09treaties in place to deal with the the types of weaponry that we're
00:23:12we're talking about here? So there
00:23:15there are, of course treaties in in place. Their ability to keep up
00:23:18up with the pace of technological technological development. That is
00:23:22is indeed another question. The Outer Outer Space Treaty is really different.
00:23:25different. Treaty of outer space space activities. However, it's also
00:23:28also the product of its time. So
00:23:30So it focuses primarily on traditional traditional warfare technologies
00:23:34technologies such as weapons of mass
00:23:35mass destruction and nuclear weapons. weapons. But it doesn't really appear
00:23:39appear to cover the deployment of of the full range of counter space
00:23:42space options that right now we are
00:23:44are seeing under development, including including those, especially that
00:23:47that are carried out via Non-kinetic
00:23:50Non-kinetic electronic or cyber means means eventually. So in that sense,
00:23:54sense, we don't have a safeguard safeguard there. We have, of course,
00:23:57course, however, and it is important important also to mention we have
00:24:00have all the safeguards that we find, find, whether international or in
00:24:03in relation to the use of force and
00:24:05and in relation to international international humanitarian law. So
00:24:09So the law during armed conflicts.
00:24:11conflicts. OK. Julia Pervizi from from the European Space Policy Institute,
00:24:15Institute, thank you very much for for your time. Now, in what's believed
00:24:18believed to be an act of sabotage, sabotage, train travel across large
00:24:22large parts of the Netherlands ground ground to a standstill today, hours
00:24:25hours before the ceremonial state state opening of Parliament services
00:24:29services in central and eastern parts parts of the country suffered severe
00:24:32severe disruption. Authorities suspect suspect this was a co-ordinated act
00:24:36act of sabotage. A statement from
00:24:38from the Dutch rail operator ProRail
00:24:40ProRail said objects had been found found attached to train tracks in
00:24:44in multiple locations. It said no
00:24:46no suspect or motive had been identified identified yet, but the National
00:24:50National Police Unit is investigating. investigating. Our correspondent
00:24:52correspondent Nick Beake is in Brussels, Brussels, where the disruption is
00:24:55is already affecting neighbouring neighbouring countries. Yes, mainly
00:24:59mainly in the north and the east east of the country where the disruption
00:25:02disruption has been seen from early early this morning, it has to be be said, and the information we've
00:25:05we've got is from the Dutch rail
00:25:07rail operator saying that there thinking
00:25:10thinking that this is a co-ordinated co-ordinated act of sabotage, they're
00:25:14they're talking about around 30 spots,
00:25:16spots, mainly in rural locations. locations. These are places where
00:25:20where it seems that someone has attached attached some sort of tube or pipe
00:25:24pipe to the track. And the problem problem is that the computer system
00:25:28system detects this as being a part part of the track that's occupied.
00:25:30occupied. So in other words, the the computer thinks that there is
00:25:34is some sort of train there at the the moment. So that kicks into place
00:25:38place a red light. And so you get
00:25:40get all this disruption along the
00:25:42the tracks. And we've seen this today,
00:25:45today, lots of cancellations, lots lots of problems. The big question
00:25:48question is who is behind this. And And we don't have an answer to that.
00:25:51that. We know that the various authorities,
00:25:54authorities, the police prosecutors prosecutors also intelligence agencies agencies in the Netherlands are looking
00:25:58looking into this. One thing to say, say, though, this is the Dutch equivalent
00:26:02equivalent of the state opening of of Parliament. It's when the King
00:26:05King gives a speech, the government government sets out its budget for
00:26:09for the next year and farmers had had said that they were going to
00:26:12to protest in some way today. They They haven't said that they're behind
00:26:15behind this, but certainly lots of of people, lots of groups in the
00:26:18the Netherlands unhappy about the the budget for the year ahead. And
00:26:21And so that will certainly be one one line of inquiry that the authorities
00:26:24authorities will be looking at. Nick Nick Beake let's return to our top
00:26:28top story now on AI. We can show show you the scene live in Washington.
00:26:31Washington. Bernie Sanders still still speaking there at the Pro-human
00:26:35Pro-human Assembly in Washington, Washington, DC we're expecting in in a moment to hear from Steve Bannon,
00:26:39Bannon, an ally of President Trump. Trump. But he is hugely critical
00:26:42critical of President Trump's reluctance reluctance to order federal regulation.
00:26:46regulation. Max Corbridge is still
00:26:48still here with me, CEO of AI safety
00:26:51safety company Secure Agent Max. Max. It was also really interesting
00:26:54interesting to hear from Bernie Sanders Sanders a few minutes ago where he
00:26:58he was talking about AI escaping
00:27:01escaping human control. We know there
00:27:03there have been instances of AI models models doing that, but if we look
00:27:07look at that on a larger scale, how how possible how likely is that briefly?
00:27:10briefly? Well, that's exactly it. it. We've seen on the small scales scales already where you give a task
00:27:14task to an AI agent and it does something something that you didn't anticipate. anticipate. And we've actually deemed
00:27:17deemed that loss of control. But But what we're really running towards
00:27:20towards is where we're using AI to to create the next frontier of AI.
00:27:23AI. And that's where we kind of zoom zoom out a bit and we see the existential
00:27:26existential risk. And that's really really something that OpenAI have have said they want to do in the
00:27:30the next two years. And it's something something that really then we put put we put the gas pedal down without
00:27:34without the ability to pull the brakes brakes up because we're no longer longer the ones creating these AI
00:27:38AI systems. Yeah. OK. Max, thank
00:27:40thank you very much for that. And And yeah, just a reminder of that
00:27:43that breaking news in the last couple couple of minutes. OpenAI Anthropic
00:27:47Anthropic and Google talking about
00:27:50about creating a guardrail, a kind kind of a way of keeping check on
00:27:54on AI, but huge questions about whether whether the industry itself should
00:27:57should be responsible for creating creating a standards body, whether
00:28:00whether it should be a body that that is independent of the industry.
00:28:03industry. We're going to keep you you right up to date with developments developments in
00:29:36still in the mid to high 40s by.