it is all change in kyiv. president zelenskiy has today fired his top army commander in the biggest shake up of the armed forces since russia's full—scale invasion nearly two years ago. general zaluzhnyi was asked to resign monday, but refused to quit. so today, he was sacked. the president issued a photograph of the two men together, thanking zaluzhnyi for his service, but outlining the need for "renewal". but the ever—popular general has not gone without a fight, and before he was officially fired, he put out a statement that will raise some serious concerns within the armed forces. our partners�* stocks of missiles, air defence interceptors and ammunition for artillery is becoming exhausted, he wrote. russia has taken note of how developments in the middle east have

distracted international attention, and will now seek, he says, to provoke further conflicts elsewhere. he pointed towards the weakness of the international sanctions regime, which means russia is still able to build and maintain a military—industrial complex in pursuit of war. and he points the finger at the draught in ukraine, which is not, he said, currerntly fit for purpose. so how will international partners interpret all of that? we arejoined by — yaroslav trofimov, the chief foreign affairs correspondent of the wall streetjournal. and the author of the recently published book — our enemies will vanish: the russian invasion and ukraine's war of independence. thank you for being with us again on the programme. so... there has been a lot reported about this in recent weeks. what is the official reason for this decision and what is the real reason?— real reason? ella, the official reason is _ real reason? ella, the official reason is that _ real reason? ella, the official reason is that ukraine - real reason? ella, the official reason is that ukraine needs| real reason? ella, the official- reason is that ukraine needs new leadership and new blood and it

needs to rethink and reboot its strategy. in the war that is now approaching a two year mark. general zaluzhnyi was immensely popular and remains immensely popular in ukraine, and he is credited with really serving the country in the first weeks of the war when ukraine was written off by its allies in the west and was expected to fall in a matter of days. he engineered a sort of nimble mobile defence, trading land for lives of soldiers and four—time, and really was instrumental in the survival of ukraine's estate. but he also had quite a long history of disagreements with president zelensky and it is no secret that the new commander of ukrainian forces was always a favourite of president zelensky.— forces was always a favourite of president zelensky. ride, so are you sa in: that president zelensky. ride, so are you saying that there _ president zelensky. ride, so are you saying that there was _ president zelensky. ride, so are you saying that there was in _ president zelensky. ride, so are you saying that there was in fact - president zelensky. ride, so are you saying that there was in fact a - saying that there was in fact a split between the army commander and the head of ground forces who is his

replacement? i the head of ground forces who is his replacement?— replacement? i wouldn't call it a slit, but replacement? i wouldn't call it a split. but if— replacement? i wouldn't call it a split, but if the _ replacement? i wouldn't call it a split, but if the decision - replacement? i wouldn't call it a split, but if the decision making| replacement? i wouldn't call it a i split, but if the decision making in the military was in general zaluzhnyi's cans in the first week of the war, by may 2022, all of main decisions were taken by the high command of the military. various commanders were all sitting together and making calls on the very important pivotal points of the war such as the defence of a specific region. it is not like general zaluzhnyi single—handedly was running the military in the first place. running the military in the first lace. ~ ., running the military in the first lace, ~ ., ., running the military in the first lace. ~ ., ., , ., place. what about the statement he ut out place. what about the statement he put out before _ place. what about the statement he put out before he _ place. what about the statement he put out before he was _ place. what about the statement he put out before he was sacked? - place. what about the statement he put out before he was sacked? it. place. what about the statement he put out before he was sacked? it is| put out before he was sacked? it is a litany of problems that he sees, how will international partners view that? i how will international partners view that? 4' , ., . ~ ., that? i think he is acknowledging the reality- _ that? i think he is acknowledging the reality. the _ that? i think he is acknowledging the reality. the war— that? i think he is acknowledging the reality. the war is _ that? i think he is acknowledging the reality. the war is at - that? i think he is acknowledging the reality. the war is at a - the reality. the war is at a stalemate if you look at the front lines, they have not moved significantly for more than a year. and ukraine is running very low on

ammunition, in part because of the shortfall in western funding and because of the us congress not moving on the defence appropriations for ukraine for a month now. and what he said is correct, about the need to invest in new technologies and develop drone warfare. that is something that people realise across the border in ukraine. one point that gets lost in all of this debate is that as part of the reshuffle today, president zelensky also promoted several battlefield commanders and brigade commanders who are legendary in some parts of the front line for their successes in battles throughout the war. such as the commander of the brigade that secured the city early in the war. and so this is all portrayed as infusing fresh blood and battle tested commanders across line into the command headquarters of the

ukrainian armed forces. i’m the command headquarters of the ukrainian armed forces.— ukrainian armed forces. i'm 'ust fascinated — ukrainian armed forces. i'm 'ust fascinated by �* ukrainian armed forces. i'm 'ust fascinated by the internal h ukrainian armed forces. i'mjust| fascinated by the internal politics of the ukraine. if you look at polling, general zaluzhnyi was as popular, if not more popular, then president zelensky. the fact that he has fired him, how does that play politically in the country? i know there properly would be elections next year, even though they were slated, but how will people view that and react to that? i slated, but how will people view that and react to that?— that and react to that? i think a lot of peeple — that and react to that? i think a lot of people really _ that and react to that? i think a lot of people really appreciate i that and react to that? i think a i lot of people really appreciate the role that general zaluzhnyi played, and i think they were looking for more explanations of why it had happened. it is generally accepted, the dismissal. we must all keep working together and focus on the common enemy. general zaluzhnyi has always denied that he harboured any political ambitions, whether that is true we will see in the coming months and years.— true we will see in the coming months and years. indeed so, now that he is out _ months and years. indeed so, now that he is out of— months and years. indeed so, now that he is out of the _ months and years. indeed so, now that he is out of the job, _ months and years. indeed so, now that he is out of the job, we - that he is out of the job, we probably will. yaroslav trofimov, always good to talk to you, thank

you for coming in the programme. around the world and across the uk. this is bbc news. bbc news, bringing you different stories from across the uk. - a couple from manchester with a special place in their hearts for cumbria. second homeowners like howard and mandy are due to get clobbered with a doubling of council tax from next april. we try to put as much as we can into the community and to be involved in the local area when we are here, so we don'tjust come and sort of bring all our stuff and go. a short drive away, don says villagers in the above derwent parish have seen a dramatic rise in second homes. and he thinks second homeowners should pay more council tax. in this parish we have a lot of holiday lets also. and this is having a knock—on effect of young people being able to afford a house to live in the area. cumberland council says the second home council tax premium would still raise £5 million, some of which could be used to tackle the housing crisis in beauty spots like this. more stories from across the uk,

head to the bbc news website. l you are watching the context. and now it is time for our our new weekly segment — ai decoded. my favourite part of the week, welcome to ai decoded, in which we deep dive into some of the most eye—catching stories in the world of artificial intellligence. let me give you a rundown of what we are covering tonight. starting with this in the mail. a team in the united states that simulated present—day war scenarios found the ai models they had tasked to run the operations, openai, meta, anthropic, all chose to escalate, right up to a nuclear response. which in itself is a case for more rules and regulation. the europeans are stealing a march in legislating ai. but reuters reports today that the us is responding with a new ai consortium drawing from 200 of the biggest companies in the us,

who they hope will drive the safe development and deployment of generative ai, which is already being deployed in indonesia's presidential election. one of the contenders in the world's third—largest democracy is the current defence minister general prabowo subianto, who has been given an ai make—over. once feared as a special forces commander, he is now some chubby—cheeked avatar that is winning the hearts and minds of millions of young voters. this is extraordinary. the guardian says hong kong police have launched an investigation after an employee at an unnamed company claimed she was duped into paying $25 million of her firm's money to fraudsters who staged a deepfake video call, in which they'd appeared as her boss. finally, swipe left for no or right for a date. how many of you spend hours on tinder or grindr, searching the perfect partner? well not aleksandr zhadan. from his apartment in moscow, he set chatgpt to work. it interviewed over 5000 women on his behalf, until it found him the perfect match

karina, who is now his wife! here is our regular host priya lakhani, ceo of century tech, an ai intelligence, education, technology company that develops ai tools. hello. good to see you. there is a lot written — hello. good to see you. there is a lot written about _ hello. good to see you. there is a lot written about al _ hello. good to see you. there is a lot written about al and _ hello. good to see you. there is a lot written about al and how - hello. good to see you. there is a lot written about al and how it - lot written about al and how it would react if it were running defence systems.— would react if it were running defence systems. would react if it were running defence s stems. , ., , defence systems. yes, the reason why there is a study _ defence systems. yes, the reason why there is a study which _ defence systems. yes, the reason why there is a study which came _ defence systems. yes, the reason why there is a study which came up - defence systems. yes, the reason why there is a study which came up today l there is a study which came up today and why it is really interesting is that it and why it is really interesting is thatitis and why it is really interesting is that it is increasing interest from governments to use artificial intelligence to potentially make decisions in the future. they conducted a study where they took, basically, what age nations may look like. we have global superpower with expansionist ambitions all the way to a newly independent state. they did inspire those based on real states. but they are fictional. so they took these eight fictional states with unique foreign—policy goals, they put them in three

scenarios. one is neutral, there is nothing happening here. the other one is there is a cyber attack. the other one was a full—scale invasion. and then with the simulations they used chat bots, but different chat bots, that we have discussed before on the market, openai, chatterjee bt, and tropics, and then they looked at how those chat bots would respond and what was really, really scary was that they escalate very quickly. and the reasons why they escalate, christian, you would be alarmed. one was, we have got them so we will use them as a deterrent. we have got a nuclear weapon so let's go nuclear really quickly. let's just strike first.- let's go nuclear really quickly. let's just strike first. let's 'ust strike first. have we got a let'sjust strike first. have we got a rah? let'sjust strike first. have we got a graph? let's— let'sjust strike first. have we got a graph? let's put— let'sjust strike first. have we got a graph? let's put it _ let'sjust strike first. have we got a graph? let's put it on _ let'sjust strike first. have we got a graph? let's put it on screen. i let'sjust strike first. have we got l a graph? let's put it on screen. you will see the invasion scenario, so i see, on the right—hand side and on the near right—hand side you see how

chatterjee bt responded, and it is really the forth one which goes nuclear quickest. {eft really the forth one which goes nuclear quickest.— really the forth one which goes nuclear quickest. gpt for bass is different to _ nuclear quickest. gpt for bass is different to gpt _ nuclear quickest. gpt for bass is different to gpt for. _ nuclear quickest. gpt for bass is different to gpt for. it _ nuclear quickest. gpt for bass is different to gpt for. it is - nuclear quickest. gpt for bass is different to gpt for. it is less . different to gpt for. it is less sophisticated but it has not got enough fine tuning, it is more the budget virgin. the black lines, the vertical black bars which look like mini dumbbells, they are error rates. some of these bars are statistically not significant. that is really important. what this actually shows the frequency at which the models, the language models are up there and at the cuddle bars, you have the key on your right hand side. it shows the frequency at which those models, in the simulations, would basically take one of the six scenarios along the access keys, the horizontal one. how often would they de—escalate? can you see how straightaway you have these situations where there... over a two—week period, there are a better chance in the report, the study that i decided to read for this, but it is actually really

interesting.— this, but it is actually really interestinu. ., , , . , interesting. maia dee instinctively escalatinu ? interesting. maia dee instinctively escalating? this _ interesting. maia dee instinctively escalating? this is _ interesting. maia dee instinctively escalating? this is all _ interesting. maia dee instinctively escalating? this is all about - interesting. maia dee instinctively escalating? this is all about how i escalating? this is all about how the are escalating? this is all about how they are trained _ escalating? this is all about how they are trained and _ escalating? this is all about how they are trained and what - escalating? this is all about how they are trained and what they i escalating? this is all about how i they are trained and what they are trained on. , ., they are trained and what they are trained on— they are trained and what they are trained on. , ., a , trained on. they are mimicking us, essentially? _ trained on. they are mimicking us, essentially? but _ trained on. they are mimicking us, essentially? but fast _ trained on. they are mimicking us, essentially? but fast forwarding i trained on. they are mimicking us, essentially? but fast forwarding it. j essentially? but fast forwarding it. there is anyone see. actually, one of the researcher stated that it could be because they have been trained on more data to do with escalation scenarios. more than diplomacy. you have got to remember, all of these models are trained with lots of data and information, and so they are going to act with... this is very simplistic, the majority of what they are trained on. i think the key point here is that anyone in the key point here is that anyone in the military who is looking at using these... , ., ,., , these... they are, the point here is that the us — these... they are, the point here is that the us military _ these... they are, the point here is that the us military is _ these... they are, the point here is that the us military is working - these... they are, the point here is that the us military is working with | that the us military is working with openai to incorporate that sort of tech into its hardware. you might guess, the point is that there is a significant difference in behaviour as you can see between the different models, the as you can see between the different models, . , as you can see between the different models, , , ., , ., ., models, the study goes into that ruite models, the study goes into that quite extensively. _ models, the study goes into that quite extensively. the _ models, the study goes into that quite extensively. the choice - models, the study goes into that quite extensively. the choice of. models, the study goes into that | quite extensively. the choice of a large language model will be really

important, but you can't trust it. so it shouldn't be an agent, the idea of ai notjust been prompted by a human but actually taking an autonomous decision by itself and then acting out. we autonomous decision by itself and then acting out.— then acting out. we have talked about drone _ then acting out. we have talked about drone was _ then acting out. we have talked about drone was on _ then acting out. we have talked about drone was on this - then acting out. we have talked - about drone was on this programme. when you think that computers might be running these drones and a armada of drones, that is a worrying aspect of drones, that is a worrying aspect of it. it brings us neatly to the story from writers where the biden administration are setting up this new consortium, so a representative from each of the 200 companies that are taking part, what is the principal? if are taking part, what is the principal?— are taking part, what is the --rincial? i. , . ,, principal? if you remember back in the ai summit _ principal? if you remember back in the ai summit where _ principal? if you remember back in the ai summit where we _ principal? if you remember back in the ai summit where we lodge - principal? if you remember back in j the ai summit where we lodge this part of the programme... you launch, ou are part of the programme... you launch, you are next — part of the programme... you launch, you are next to _ part of the programme... you launch, you are next to rishi _ part of the programme... you launch, you are next to rishi sunak_ part of the programme... you launch, you are next to rishi sunak doing - you are next to rishi sunak doing it. i you are next to rishi sunak doing it. ., , ., , ., , it. i was, and when the summit was takin: it. i was, and when the summit was taking place — it. i was, and when the summit was taking place at _ it. i was, and when the summit was taking place at bletchley _ it. i was, and when the summit was taking place at bletchley park, - taking place at bletchley park, there was a us executive order by president biden and kamal harris as he and she announced the executive order. as part of that executive order. as part of that executive order which is the us's approaching to how we are going to move forward with regulation and safeguarding

artificial intelligence technology, they asked mr, which is the national institute of standards and technology in the us, to create an institute of ai safety. as part of that institute, they will create this consortium. and this consortium will have members and representatives of academia, government, industry, all together so that they stand the best chance of being able to create the best guidelines and policy going forward to safeguard against risks. that guidelines and policy going forward to safeguard against risks.- to safeguard against risks. that is where i to safeguard against risks. that is where i get _ to safeguard against risks. that is where i get concerned, _ to safeguard against risks. that is where i get concerned, we - to safeguard against risks. that is where i get concerned, we have i to safeguard against risks. that is . where i get concerned, we have been spoken about this a lot, he goes into congress and puts the proposals and directives to them and they don't understand them. by the time they understand it, the technology has already moved on. so can you be led by a consortium, or do you have to legislate on each and everything? that is a classic case of how regulation always chases innovation. that happens all of the time. so the consortium won't lead, what it will do is it will help. and it will provide... 600 entities, i believe, applied for this, 200 were selected finalists to be part of the

consortium. there is criteria, you have to be able to add expertise, offer data and models etc. really interesting point about the us is that the government has been criticised that mist is really underfunded. so what has not been decided is how much funding the safety institute will get and the consortium have in order to do this. one of the key aspects they want them to do is create this red teaming framework.- them to do is create this red teaming framework. them to do is create this red teamin: framework. ~ . , . teaming framework. what is that? or red teaming — teaming framework. what is that? or red teaming is _ teaming framework. what is that? or red teaming is the _ teaming framework. what is that? or red teaming is the idea _ teaming framework. what is that? or red teaming is the idea that - teaming framework. what is that? or red teaming is the idea that you - red teaming is the idea that you have a team of people, ethical hackers, analysts... you test the systems? hackers, analysts... you test the s stems? ., ., ., ,, , hackers, analysts... you test the s stems? ., ., ., _ , ., systems? you go into the system and emulate what — systems? you go into the system and emulate what an _ systems? you go into the system and emulate what an adversarial- systems? you go into the system and emulate what an adversarial attack. emulate what an adversarial attack would look like. so when you do that, it is really, really popular when it comes to cybersecurity. lots of people use red teaming. you are able to uncover inefficiencies or weaknesses of and abilities of the system. the us have a history of doing this, so they have big pentagon hack day in 2016 where there was a generative ai hacker

afon at def con last summer. so what they are going to do is create a framework for red teaming, which is why it is important to have all of the right sort of voices at the table. these are different voices. you have a big large language model created and operated but it goes beyond. created and operated but it goes be ond. ~ , ., ., beyond. when they do it on the defence systems, _ beyond. when they do it on the defence systems, they - beyond. when they do it on the defence systems, they will - beyond. when they do it on the defence systems, they will find j beyond. when they do it on the - defence systems, they will find that they are helping to nuclear war, so maybe they will wrecked in the defence systems. let's talk about the story, since we are talking about generative ai, in indonesia. the pot has always been, the concern, with an election is that there will be defects and misinformation. what we have not really looked at on a programme is how generative ai might be used to help candidates.— help candidates. yes, and i think renerall help candidates. yes, and i think generally across _ help candidates. yes, and i think generally across many _ help candidates. yes, and i thinkj generally across many countries, there is a stance that it should not. it should not help in political campaigning. i? because democracy is all about individuals having informed choices and decisions on a candidate, what they are saying, what they are doing. if you look at the image which has been created... let's bring it up. this is the general who is running. this

let's bring it up. this is the general who is running. generalwho is running. this is a reneral generalwho is running. this is a general who _ generalwho is running. this is a general who has _ generalwho is running. this is a general who has formally - generalwho is running. this is a general who has formally are . general who is running. this is a i general who has formally are quite hard character. but he has been depicted here using generative ai, this is his team and his campaign, he is the sort of cuddly ai avatar. what is fascinating is that 235 million voters in indonesia, half of them are under a0 years old and it turns out that this sort of cuddly, cute character is really appealing to the sort of genesee and millennial.— to the sort of genesee and millennial. ~ , , . ., millennial. door kicking special services general _ millennial. door kicking special services general is _ millennial. door kicking special services general is depicted - millennial. door kicking special services general is depicted as| millennial. door kicking special. services general is depicted as an avatar, cuddly avatar, and suddenly he is rising at the polls! i5 avatar, cuddly avatar, and suddenly he is rising at the polls!— he is rising at the polls! is rising at the polls- _ he is rising at the polls! is rising at the polls. there _ he is rising at the polls! is rising at the polls. there has - he is rising at the polls! is rising at the polls. there has been - he is rising at the polls! is rising - at the polls. there has been decades of research in the psychology and it what it shows, i really like this, great article that viewers can look up, images can produce erroneous memories. something called the... a specific effect where you can have a more substantive claim by showing an image. if you have this political

campaign and i should be soft and cute images, saying on a billboard soonin cute images, saying on a billboard soon in the uk,...— cute images, saying on a billboard soon in the uk,... rishi sunak, keir starmer--- — soon in the uk,... rishi sunak, keir starmer. .. prime _ soon in the uk,... rishi sunak, keir starmer. .. prime minister - soon in the uk,... rishi sunak, keir starmer. .. prime minister cuddly, l starmer. .. prime minister cuddly, challenae starmer. .. prime minister cuddly, challenge cute, _ starmer. .. prime minister cuddly, challenge cute, i— starmer. .. prime minister cuddly, challenge cute, i don't _ starmer. .. prime minister cuddly, challenge cute, i don't know. - starmer. .. prime minister cuddly, | challenge cute, i don't know. they should not now say that i'm calling them cute and cuddly, but the point is... i should them cute and cuddly, but the point is... ishould not them cute and cuddly, but the point is... i should not have said that. that will haunt me forever. flan is... i should not have said that. that will haunt me forever. can we edit that out? — that will haunt me forever. can we edit that out? | — that will haunt me forever. can we edit that out? i know _ that will haunt me forever. can we edit that out? i know it _ that will haunt me forever. can we edit that out? i know it is - that will haunt me forever. can we edit that out? i know it is like. - edit that out? i know it is like. you can't, it is like. i know you are interested, so i use the same software, it took me five minutes and that is you. the wonderful karen who does your hair and make—up, by the way, said the first one really looks like christian. and she said it looks lovely and cute and cuddly. what, in real life? but it looks lovely and cute and cuddly. what, in real life?— what, in real life? but can you see how obviously _ what, in real life? but can you see how obviously that _ what, in real life? but can you see how obviously that is _ what, in real life? but can you see how obviously that is fun, - what, in real life? but can you see how obviously that is fun, but - what, in real life? but can you see| how obviously that is fun, but when thatis how obviously that is fun, but when that is on billboards all over indonesia, if it is on billboards it creates a different persona. you have created your persona rather than actually people are looking at you, right, in your normalform and thinking about the policies and what you are saying. so it can skew the

way in which people vote. i thought u . way in which people vote. i thought u- and ai way in which people vote. i thought up and al had _ way in which people vote. i thought up and ai had stopped _ way in which people vote. i thought up and ai had stopped candidates l up and ai had stopped candidates from using generative ai.- from using generative ai. openai has, from using generative ai. openai has. right. _ from using generative ai. openai has. right. but— from using generative ai. openai has, right, but many— from using generative ai. openai has, right, but many platforms i from using generative ai. openai i has, right, but many platforms have policy saying you are not allowed to use... this is mid—journey, mid—journey have also said you are not allowed to use images for political campaigns. pare not allowed to use images for political campaigns.— political campaigns. are your twitter feed? _ political campaigns. are your twitter feed? exactly. - political campaigns. are your twitter feed? exactly. they l political campaigns. are your- twitter feed? exactly. they have come a bit _ twitter feed? exactly. they have come a bit also _ twitter feed? exactly. they have come a bit also raises _ twitter feed? exactly. they have come a bit also raises another i come a bit also raises another point. if you are violating and infringing the terms and conditions of the platform and creating these things for campaigning purposes, i was doing it because it is educational for everyone on the news, but the point is if you are doing that, then what is the consequence? and this is the issue. coming to regulation, there needs to be a hurry. hate coming to regulation, there needs to be a hur . ~ ., ~ coming to regulation, there needs to beahur .~ .,~ be a hurry. we were talking about how ou be a hurry. we were talking about how you might — be a hurry. we were talking about how you might misrepresent - be a hurry. we were talking about - how you might misrepresent yourself, this story in hong kong is i think they're a story of the day because itjust has they're a story of the day because it just has so they're a story of the day because itjust has so many implications for us. let me tell the story again. this woman is having... cheese on a

video conference call, the boss, or the office manager, appears to her virtually, and it is not him, it is the hackers who effectively are fishing. and so they persuade her to transfer $20 million into their account... the implications of this are so profound, because we are talking about al speeding up the way we work, but if you can't trust anything, if you have to check everything and everyone, everyone that you are talking to, then we start going backwards.- that you are talking to, then we start going backwards. exactly, and when they have _ start going backwards. exactly, and when they have biometrics - start going backwards. exactly, and i when they have biometrics increasing security, you have voice authentication in the bank, for example, it creates a huge problem. there are three ways in which you can hack... lots of ways in which one can hack biometrics but the idea of replay attacks, when they essentially steal your identity and use it to con someone is a very common one. you have spoofing and

skimming, skimming is when they copy your fingerprints skimming, skimming is when they copy yourfingerprints are an skimming, skimming is when they copy your fingerprints are an atm machine, for example. do you mind if i play something for you? we know each other, we are friends, so i trust you. i created this in one minute, tops. hi, priya. it's christian, ijust need to grab a picture of your passport and also your driver's license so that the bbc security can let you in, as we have some new security protocols in place. please can you send them over to this number. thanks, see you later this evening on the context. nowhere! it does sound like you. 60 seconds tops- _ nowhere! it does sound like you. 60 seconds tops- i _ nowhere! it does sound like you. 60 seconds tops. i used _ nowhere! it does sound like you. 60 seconds tops. i used 11 _ nowhere! it does sound like you. 60 seconds tops. i used 11 alarms, - nowhere! it does sound like you. 60 seconds tops. i used 11 alarms, and. seconds tops. i used 11 alarms, and ayo. i knew you would not get me into trouble because i asked your permission to use your voice, but educational purposes again. the point is is that you now need... it is the point on trust, you now need... you can't only rely on voice... i need... you can't only rely on voice- - -_ need... you can't only rely on voice... ~ ., ., ., ., voice... i know what you are going to sa , voice... i know what you are going to say. hang _ voice... i know what you are going to say. hang on. _ voice... i know what you are going to say, hang on, on _ voice. .. i know what you are going to say, hang on, on my— voice... i know what you are going to say, hang on, on my santander| to say, hang on, on my santander account when i ring up and i go

through the process, they asked me to say a phrase because they use my voice for recognition. i to say a phrase because they use my voice for recognition.— voice for recognition. i don't know, do they do — voice for recognition. i don't know, do they do that _ voice for recognition. i don't know, do they do that now? _ voice for recognition. i don't know, do they do that now? i _ voice for recognition. i don't know, do they do that now? i know - voice for recognition. i don't know, do they do that now? i know that i voice for recognition. i don't know, i do they do that now? i know that my bank used to do that sometime ago but that is obviously a problem because your voice is available everywhere. i picked that up of some youtube video. that is an issue, so multifactor authentication is probably where this is going to go, but for the regular person when you are thinking, hang on, i have lost my phone and are they asking me to transfer money, the point is, try and verify what is going on. you know the times when you have sat on your phone... ii know the times when you have sat on your phone- - -— your phone... if someone appears to ou in your phone... if someone appears to you in screen — your phone... if someone appears to you in screen as _ your phone. .. if someone appears to you in screen as your— your phone... if someone appears to you in screen as your boss... - your phone... if someone appears to you in screen as your boss... you - you in screen as your boss... you call them- _ you in screen as your boss... you call them. use _ you in screen as your boss... you call them. use a _ you in screen as your boss... ym. call them. use a different device, use a different way of contacting them. that is the only way to do it. they may have worked very hard on the video, the image. that they may have worked very hard on the video, the image.— the video, the image. that will only aet the video, the image. that will only get better? — the video, the image. that will only get better? yes. _ the video, the image. that will only get better? yes. you _ the video, the image. that will only get better? yes. you have - the video, the image. that will only get better? yes. you have to - the video, the image. that will only get better? yes. you have to find i get better? yes. you have to find more than _ get better? yes. you have to find more than one _ get better? yes. you have to find more than one way _ get better? yes. you have to find more than one way to _ get better? yes. you have to find more than one way to verify - more than one way to verify something. the most important thing is how many times have you looked at your or computer and up ignored the

update because it sets you back? don't do that, because the reason why. .. don't do that, because the reason why... i don't do that, because the reason wh ., �* don't do that, because the reason wh .,�* ~' ., don't do that, because the reason why... i don't know if! should have 'ust said why... i don't know if! should have just said that _ why... i don't know if! should have just said that on _ why... i don't know if! should have just said that on live _ why... i don't know if! should have just said that on live television - just said that on live television but i will start updating. the point is is that we... _ but i will start updating. the point is is that we... people _ but i will start updating. the point is is that we... people use - but i will start updating. the point is is that we... people use ai - but i will start updating. the point is is that we... people use al to i is is that we... people use a! to actually stop these attacks from happening, and ai actually stop these attacks from happening, and al to look at for example networks or anomalies in traffic, vulnerabilities in the network, they are using ai traffic, vulnerabilities in the network, they are using al to combat some of the issues created by ai. but every time they send you those updates it is because they want to updates it is because they want to update your software, so that it is updated with the latest cyber security. so i'm just saying this to viewers. stop ignoring it, because i used to. but i know not to now. verifying people and who they are, let's finish with this nice story from moscow, actually. this guy trade chatgpt to be him and instead of going on the dating sites and swiping right or left, he got ai of going on the dating sites and swiping right or left, he got al to do it. . swiping right or left, he got al to do it. , , ~ swiping right or left, he got al to do it. , , ,, ., ., do it. yes, he sounds like one of the old aunties _

do it. yes, he sounds like one of the old aunties from _ do it. yes, he sounds like one of the old aunties from the - do it. yes, he sounds like one of the old aunties from the old - do it. yes, he sounds like one of. the old aunties from the old indian generation that used to do arranged marriages. when somebody wasn't married by a certain age, especially a girl, a female, they would say... everyone would laugh and say you are going to go into anti's database, she will try to match up with someone. in the olden days, they would have often have scenarios in villages where you go into the room and the couple don't even talk. the aunties do all of the talking. at least here, you have a! doing it for you. all i can say is congratulations to the happy couple. they are getting married, the person that he chose explanation or you or i don't recommend this, guys, girls, everybody, whichever gender. you are acuttin the everybody, whichever gender. you are putting the female _ everybody, whichever gender. you are putting the female chat _ everybody, whichever gender. you are putting the female chat and _ everybody, whichever gender. you are putting the female chat and personal i putting the female chat and personal data into a large language model, surely that is breaching privacy. this has got really big implications, but i don't want to take away the happiness of this wonderful couple.— take away the happiness of this wonderful couple. where it works in a more positive _ wonderful couple. where it works in a more positive way, _ wonderful couple. where it works in a more positive way, take _ wonderful couple. where it works in a more positive way, take it - wonderful couple. where it works in a more positive way, take it away i a more positive way, take it away from the dating sites. we did this story about four or five weeks ago

where someone who offered tourism advice, and he went to the golf course, the clients what they were talking to him because it was actually given the responses that he had trained to give, and so i did actually give him time away from the job. and it could work 20 a7. that is how job. and it could work 20 47. that is how companies _ job. and it could work 20 47. that is how companies are _ job. and it could work 20 a7. trust is how companies are creating efficiencies with this. you can train our language model on your website, all of the information on your website, and create a nice chat bot faq so that you don't have to hire people for customer service. using this context, i think some people at the end of the tender app might feel a little bit differently, but there are good use cases and there are bad, but hey, congrats to this man. i can't remember his name. thats it it, we are out of time. iam not i am not sending you my passport or driving licence details, and i will update my computer. we will do all this again at the same time next week. you can catch this programme

on youtube as well and we will start developing it over the last week because a lot of you are reacting to it and want to see more of it. so stay tuned. we will certainly do that in the weeks ahead. coming up after the break we will talk to our panel and we will focus on that decision pending in the supreme court, can donald trump appear on the colorado ballot? stay with us. hello. well, as predicted, the winter wonderland did arrive in the areas where it was forecast. so mostly the northern hills of wales, northern england, some in northern ireland and scotland, too. here's a picture from wakefield, beautiful snowflakes there, but elsewhere rain, heavy rain. and these are the stratus and the nimbostratus rain—bearing clouds shrouding the tops of the skyscrapers in london. ok, let's have a look at the radar, then. where it's blue, it's raining. this was earlier on. where it's white and grey, that's sleet and snow, so that weather front is moving northwards into scotland.

now another clump of rain heading towards the south of england, and warnings from the met office for snow and ice are still valid across parts of the country. so you can check that online. so the forecast through the night, particularly heavy rain moving in from the south, moving northwards. could be a0—plus millimetres of rain in some areas. across the north and the north—east there, i think it's that mixture of sleet and snow, particularly across the higher ground. but that thaw is setting in, a rapid thaw from the south. you can see 11 degrees in london in the morning, and further north about a—5 celsius. so the weather map for friday shows the low pressure still close by. here's the weather front. north of the weather front, it's still pretty chilly. that easterly wind drawing in the colder air from scandinavia. here we have that west—south—westerly, so that mild air coming in. and i think again across the highlands and grampians, further snow to come on friday, again met office warnings for that. the rest of the country, it's a case of often cloudy weather, showers continuing through the day and into the evening hours as well. not a pleasant picture for some

of us on friday and friday night. that takes us into saturday. the low pressure still with us. that weather front still there affecting the north—east of scotland, but already you can see turning milder. nine degrees in glasgow, seven in aberdeen. and the low is still with us on sunday, but this clump of rain i think raises the north—east of england and also eastern scotland for a time, but out towards the west and the south, i think the weather's going to be drier and brighter with just a scattering of showers. now, the outlook through the weekend into next week, it looks as though things could settle down a little bit as we head into monday and eventually tuesday with high pressure building in, at least for a time. bye— bye.

hello, i'm christian fraser. you're watching the context on bbc news. it's a disgrace to this country, they worked together-

with the justice department and the white house — - you're not supposed to do that. every one of these cases you see comes out of the white house, i it comes out of biden. there are definitions involved about whether the president is an officer of the united states, and whether or not a state can execute this clause, or whether it has to be congress that does it. i think it's just so outrageous that trump continues to think that he is above the law, above the constitution, and above the court system. joining me on the panel tonight is dr leslie vinjamuri, the director of the us and americas programme at the chatham house and professor of international relations at soas university of london — and doug heye, the former communications director of the republican national committee. first though, the latest bbc news.

the eu's climate service says that for the first time,