much more on the middle east and all the other stories of the day. sport — and for a full round—up, from the bbc sport centre, here's tulsen tullet. hello, from the bbc sport centre. there are two matches in the premier league at present — with manchester united, who went into the match 15th in the table up, against bottom of the ladder southampton. it's currently 0—0 at old trafford. garnacho had the best chance so far of putting it wide, currently 0—0. while in the earlier kick—off, ipswich town are looking to draw further away from the relegation zone. they�* re hosting brighton. that one is goalless with the second half about to be under way. more on the bbc sport website. emma raducanu is through to the third round of

the australian open, after a straight—sets win over amanda anisimova. the 22—year—old, who's currently world number 61, won 6—3, 7—5 after fighting back from breaks down in each set. next up, she'll face polish second seed iga swiatek in the last 32 — and it's a match the briton is very much looking forward to. i'm going to enjoy playing in a great atmosphere. i think the last two matches i've had, the crowd's been amazing, the support here, i've really felt at home. it'sjust support here, i've really felt at home. it's just been a really nice feeling to be a part of. so i'm looking forward to going out there and testing my game against the best, because ultimately, you play tennis and you look for these matches. so it'll be a great buzz of adrenaline. three—time finalist daniil medvedev has been knocked out of the australian open in melbourne by american teenager learner tien. the match which finished just before 3am local time, as the 19—year—old qualifier — who's ranked 121 in the world — won in five sets.

before this week's tournament, tien had never won a grand slam match — but he'll now face france's corentin moutet in the third round. rory mcilroy is five shots off the pace after the opening round of the dubai desert classic, as he bids to win the title for the third successive time. the northern irishman chipped in on the par three seventh hole as he finished the round on two under, with a group of three players out in front on five under par, including us amateur david ford. the 22—year—old is the number six ranked amateur in the world, and shots like this have him well—placed. in the coming hours, england take on australia in the third of their one day international matches in the multi—format ashes series. the match in hobart will see england facing an uphill battle to regain the ashes if they fail to win. nat sciver brunt is just hopeful that the players can show their true form

in the contest. i think it's important obviously, in the context of the series. but i think it's more about ourselves and giving the best showing of ourselves, and like i said, being willing to fail in the right way. if i, i guess, judge myself on the way i've played, am i happy her with that and has it stuck to the team philosophy? then yeah, i'll be happy with the result, whatever dashcam i happier?. and that's all the sport for now. you are watching the context. it is time for al decoded. welcome to ai decoded — that time of the week when we look in depth at some of the most eye—catching stories in the world of artificial intellligence. this week — britain took centre stage, as prime minister keir starmer announced his uk ai

action plan on monday as part of labour's goal to make the country a "world leader" in artificial intelligence. britain's guardian newspaper says artificial intelligence will be "mainlined into the veins" of the nation, with a multibillion—pound investment in the uk's computing capacity — this comes despite widespread public fear about the effect ai could have on their livelihoods. meanwhile, the telegraph says more than a nuclear power station's worth of electricity will be needed to meet surging demand for al in the uk, and will more than treble the industry's demands on the energy grid in the coming years. in the us, the pentagon is investing millions to bolster its deepfake detection capability, with the help of london—based smart home company hive. in an exclusive for al decoded, the company will be unveiling a pretty shocking threat report predicting 2025 will be the year of "deepfakes—gone—wild."

hive�*s new deepfake detector technology will help the pentagon prevent us adversaries from using deepfakes "for their deception, fraud, disinformation and other malicious operations." we'll be speaking to their ceo later in the programme. with me is our regular ai contributor priya lakhani, ceo of ai education company century tech — and joining us from the us, we have miles taylor, ex—us national security official under the previous trump administration. great guests to have with us on ai decoded. when we heard that first announcement from keir starmer on monday, bold talk of the future and what it means for this country, saying how

it'll bolster growth, turbo—charge growth, it will be used in everything from schools, hospitals, and detecting potholes. a different type of talk from many previous governments where the emphasis was more on the fears of ai, but this was about the possibilities and opportunities? possibilities and ouortunities? . �* opportunities? which i'm so excited about, _ opportunities? which i'm so excited about, being - opportunities? which i'm so excited about, being the - opportunities? which i'm so l excited about, being the head of an ai education company. there are some challenges to this of course, but i was really excited and i would say the tech community is very excited, when you read those recommendations, they look like a very long to do list, but i'm pleased that they seem time bound. also what the prime minister was talking about was an end to end ai plan — he talked about infrastructure, but then planning challenges. if you want to build data centres, thus the first thing you'll have to tackle. overall, they did talk about the safeguards, there's been lots on the online safety act this week which i think is interesting, and we have a

treat for viewers because our colleague has got an exclusive interview with our secretary of state, peter kyle, secretary of state, peter kyle, secretary of state for the department for science, innovation and technology. let's look at that interview and see what he says. power. ai uses a lot of power — something they're no strangers to here at culham, in oxfordshire, home of the uk atomic energy authority. ahead of the government's ai strategy launch, science and technology secretary peter kyle was shown the latest research into nuclearfusion. scientists predict fusion could one day produce near limitless supplies of clean energy — it's still in development, but ideal for energy—hungry data centres necessary to power a world filled with al. last summer, the government tasked ai adviser matt clifford with creating a uk ai action plan. the plan includes 50 recommendations, all of which are now being implemented. some of the standouts include an energy council, led by peter kyle and energy secretary ed miliband,

to address costs and solutions. a national data library, with the promise of secure access to public sector data, and support for uk—based ai companies and improved computing capabilities. and this empty field, in the atomic energy authority site, is part of a plan to partner with a private firm to build a data centre here. a pilot scheme for al growth zones across the uk, these zones offering enhanced power access and streamlined planning permissions. the power—hungry demands of ai are just one thing in a long list of challenges facing the technology. and when we consider that some of the biggest players in al are based in the united states, how does the uk ensure it has a significant voice when it comes to the development of the technology? a voice which isn't drowned out by political, regulatory, or commercial concerns across the atlantic? now, at the moment, we don't

have any cutting—edge companies that are british—owned. we have deepmind, which started in britain, but is now american—owned. now, we want to keep all of those ingredients that enable that kind of scale of innovation and investment to exist in britain. when it comes to regulation, are big tech companies and the ai industry, is it getting a bit of an easy ride from this government? no, it's absolutely not. what i've done is tasked each regulator to make an assessment of how ai will impact the sectors that they regulate, and then, to acclimatise and adapt accordingly going forward. so, each regulator needs to integrate ai into the way it regulates the sectors they are responsible for. mark zuckerberg, ceo of meta — which of course owns facebook, whatsapp, instagram — he said recently... we're going to work - with president trump to push back on governments around the world that are going - after american companies i and pushing to censor more. how can the uk effectively

regulate the development of ai, and also the wider tech industry, against companies that are taking that kind of a stance? now, we have the online safety act that has obligations on companies that want to come to britain and offer services to british citizens to keep people safe, and make sure that safety is baked into products. if meta falls foul of the online safety act, what will happen? every company that operates within britain must adapt and adhere to the law. every company — not negotiable. research into fusion energy here at culham continues demonstrating uk innovation. but will the government's ai action plan result in a home—grown company that can challenge the global ai power players? marc cieslak, bbc news. so as we mentioned here, we are

also talking to miles taylor, former us national security official under the previous trump administration. miles, i know that priya has a tonne of questions for you, so off you go. questions for you, so off you io, ~ , �* , questions for you, so off you go. miles, it's me again! we only have — go. miles, it's me again! we only have six _ go. miles, it's me again! we only have six minutes - go. miles, it's me again! we only have six minutes and i l only have six minutes and i have so many questions for you. donald trump has said he'll repeal president biden�*s executive order on al — what can we expect when it comes to ai policy from the us once trump is president in a week? is waiting with baited breath to see — is waiting with baited breath to see what those details are. but i — to see what those details are. but i think what's been made clear— but i think what's been made clear by— but i think what's been made clear by trump and team is that if they— clear by trump and team is that if they are — clear by trump and team is that if they are going to replace biden's_ if they are going to replace biden's ai executive order with something, it'll be something in regulating the technology, lowering the barriers, and enable _ lowering the barriers, and enable us ai growth across the

public— enable us ai growth across the public sector and the us, and deemphasize any focus the biden administration put on preventing harms and trust and safety, — preventing harms and trust and safety, and those sorts of things _ safety, and those sorts of things. you'll see this at donald _ things. you'll see this at donald trump is like inauguration, all these leaders of major— inauguration, all these leaders of major tech companies and ai companies sitting up and sending a signal that this administration and the tech sector— administration and the tech sector are closer than ever as popular— sector are closer than ever as popular how much does the us expect— popular how much does the us expect to — popular how much does the us expect to work alongside other countries — expect to work alongside other countries like the uk and europe _ countries like the uk and europe in regulation so different goes back a business needs to operate at a scale across the globe, so dealing with regulations from the us to the uk, then the eu, which is far more prescriptive, how did they expect businesses to be able to scale and have that global reach when there are a multitude of regulations and policies to have to adhere to? it's a great question, it's one of the —

it's a great question, it's one of the things that the vaccine technology sector and frankly the bite — technology sector and frankly the bite administration hands over— the bite administration hands over to — the bite administration hands over to incoming the bite administration hands overto incoming trump over to incoming trump administration overto incoming trump administration a question here that has— administration a question here that has not been resolved. however, i will say there is a very— however, i will say there is a very good _ however, i will say there is a very good road map for us to look— very good road map for us to look at — very good road map for us to look at how this can be done. and _ look at how this can be done. and really, it's in the past ten — and really, it's in the past ten years, what happened with the sociat— ten years, what happened with the social media companies around _ the social media companies around the world and the varying _ around the world and the varying regulations on social media — varying regulations on social media companies, they provide a road _ media companies, they provide a road map— media companies, they provide a road map to how this might be approached — which frankly starts _ approached — which frankly starts with a lot of big negotiations at the top trying to get— negotiations at the top trying to get governments to align on principles that then feed down into regulatory schemes that can work— into regulatory schemes that can work together well. but that'ii — can work together well. but that'll be a much taller order on al — that'll be a much taller order on al then it was with social media — on al then it was with social media platforms because this will affect every single industry. so the incoming trump teami _ industry. so the incoming trump team, especially those working on foreign policy, are very focused _ on foreign policy, are very focused on this being one of the top _ focused on this being one of the top items in the agenda, and — the top items in the agenda, and i— the top items in the agenda, and i suspect you'll hear about that in— and i suspect you'll hear about that in trump is 's inaugural address _ that in trump is 's inaugural address. .,

that in trump is 's inaugural address— that in trump is 's inaugural address. ., ., _ ., ~ address. you obviously worked for the former _ address. you obviously worked for the former trump - for the former trump administration, and i know you have a new department of efficiency in the usa, headed by a certain tech tycoon, and i know you pay attention to the uk, our chancellor has just made it more expensive for employers to have employees. and in this particular plan, they were talking about productivity, productivity, productivity. but when elon musk went into x, he got rid of a lot of staff and is very focused on efficiency, efficiency, efficiency. so people might be worried about this bold ai action plan, and that might be asking, while we know ai has lots of opportunities, what about me and my livelihood? we know it takes a long time to upscale people and gained from those newjobs people and gained from those new jobs that people and gained from those newjobs that might be created. what do you think the new president will do in a week's time when it comes to upscaling people, and how would you give people, and how would you give people the confidence that you have them in mind and it's not all about cuts, redundancies, and efficiencies? i all about cuts, redundancies,

and efficiencies?— and efficiencies? i think you hit on the — and efficiencies? i think you hit on the operative - and efficiencies? i think you hit on the operative word, l hit on the operative word, upscaling — how do we get populations adapting ai into the workplace and doing so rapidly— the workplace and doing so rapidly to understand how it will improve productivity, but also — will improve productivity, but also improve quality of life? i think— also improve quality of life? i think that's a place where you're _ think that's a place where you're likely to see the incoming administration have more — incoming administration have more in — incoming administration have more in common with the prime minister— more in common with the prime minister and his plan than he will might think. there is an effort — will might think. there is an effort to— will might think. there is an effort to try to tell the story of a — effort to try to tell the story of a democrat ai deregulation -- ai — of a democrat ai deregulation —— ai deregulation, it'll make construction cheaper. but what people — construction cheaper. but what people don't see yet, both in terms — people don't see yet, both in terms of— people don't see yet, both in terms of the upside and the downsides, is that we are only in the — downsides, is that we are only in the first— downsides, is that we are only in the first innings of the ai erai — in the first innings of the ai era. and— in the first innings of the ai era, and most people think of ai, era, and most people think of at, just — era, and most people think of at, just as— era, and most people think of ai, just as chatgpt, they put in a — ai, just as chatgpt, they put in a question and get this magical— in a question and get this magical answer that comes quickly _ magical answer that comes quickly. in 2025—26, we will see — quickly. in 2025—26, we will see a — quickly. in 2025—26, we will see a lot _ quickly. in 2025—26, we will see a lot of these technologies become — see a lot of these technologies become a genentech — in other wordsi _ become a genentech — in other words, they'll do tasks for you

like book— words, they'll do tasks for you like book plane tickets and book— like book plane tickets and book hair appointments — but what — book hair appointments — but what wiii— book hair appointments — but what will blow people's mines as you'll— what will blow people's mines as you'll see signalling about companies adapting ai into robotic— companies adapting ai into robotic systems that people can bring _ robotic systems that people can bring into— robotic systems that people can bring into their homes, to full iaundry— bring into their homes, to full laundry and help with chores. that'ii— laundry and help with chores. that'll get people very excited about — that'll get people very excited about the productivity gains, about — about the productivity gains, about how ai will change the world, — about how ai will change the world, but it'll be tough to escape _ world, but it'll be tough to escape the image of the terminator. people have a hard time _ terminator. people have a hard time not — terminator. people have a hard time not seeing how this will be used _ time not seeing how this will be used for nefarious purposes, and that's— be used for nefarious purposes, and that's where government seem — and that's where government seem to _ and that's where government seem to work together. and we'll be talking _ seem to work together. fific we'll be talking about that later on. that's a word i've never heard before but it's obviously the word for 2025. laughter you learn a new word every week in this programme. coming up on al decoded — we'll be speaking to hive ceo kevin guo. his company is currently leading the us pentagon's deepfake detection efforts, and they are just about to release release a bombshell report on al deepfakes threats for 2025. around the world and across

the uk, this is bbc news.

welcome back to ai decoded. as we mentioned before the break, 2025 could be the year of artificial intelligence deepfakes—going—wild. our next guest is warning bad actors around the world are looking to use al to disrupt politics, spread disinformation, create elaborate scams, and spread propaganda which could lead to panic. before we speak to the ceo of hive, kevin guo, let's take a look at an example — this ai—generated deep fake of microsoft boss bill gates. it can be done in a way that none of those failures of the past would recur, because just the physics of how it's built... this is deepfake example of

what is possible with powerful computer and editing. it took around 72 hours to create this example from scratch, using extremely powerful gpu. it could improve with more computing time, but 90% people cannot tell the difference. 153