longer

The world’s leading expert in identifying deepfakes has a big problem. No longer able to identify them

by

Is called Hany Farid and is considered one of the world’s leading experts on deepfake videos. This digital forensics expert was capable of uncovering videos manipulated by governments, for example, but now he has decided to leave Silicon Valley for one simple reason: It is no longer able to differentiate the real ones from those that are being generated with AI tools. And we are not surprised. Deepfakes indistinguishable from reality. In the last two decades Farid, 60, has specialized in identifying fake videos. This professor at the University of California at Berkeley has confessed that advances in generative AI have made traditional detection methods are no longer of any use. Their conclusions confirm the feeling that we have had for a long time with this type of content: AI has advanced so much that the problem is no longer just deepfakes: it is that we distrust even real photos. Farid’s reputation precedes him. His father worked for 50 years as a chemist at Eastman Kodak, and Farid grew up visiting the dark room often, watching photos become photos as they passed through the different liquids. He ended up designing a “digital fingerprint” system that made it possible to detect cases of child pornography hidden on the Internet. In fact, its technology has led to 30 million cases of potential abuse being reported each year, as well as leading to hundreds of arrests and several rescues. I surrender. Faced with the avalanche of perfect deepfakes generated with AI, Farid has decided to leave his job to take refuge on a farm in Vermont. His surrender is the latest demonstration of a harsh reality: We can no longer trust what we see on networks. Now he is dedicated to working with wood, and has distanced himself from networks and technology. The missile that changed everything. The turning point that demonstrates this crisis of this digital forensic task occurred after the viral spread of a video showing the alleged impact of a US missile on a school in Iran. Farid spent an entire day breaking down the sequence frame by frame: analyzing the geometry of the shadows, the sound delay rate of the explosion according to the laws of physics, or the pixel length of the projectile. Impossible to decide if it is false or not. He found nothing that could prove that the video was fake, and the same thing happened to other specialists. None could issue a clear verdict of authenticity, and that made it clear that AI video generation is currently so advanced that real content is indistinguishable from a deepfake generated with these latest generation models. Verifying is too complicated. There is another problem here: generating a fake video, whether toxic or not, with cloned voices that are perfectly synchronized with the interlocutor is easy, fast and cheap. Carrying out a forensic investigation to try to detect whether the video is real or not takes hours of computational and direct analysis by specialists. Given that deepfakes manage to go viral in just 20 minutes if they are successful, the methods to contain this spread are useless for a simple reason: they arrive late. The biter bit. The researcher himself was a victim of this reality: cybercriminals cloned his phone number and used AI to generate his voice and thus impersonate his identity. With that clone, they called a close contact who was involved in a court case and managed to extract confidential information. Farid and his wife, a vision researcher at Berkeley, they had to create a secret safe word at the beginning of each family call to certify that each interlocutor was who they said they were. The situation generates a disturbing paranoia and mistrust. “I’m going blind”. In the report of The New York TimesFarid explained that his studies show that most people can no longer differentiate a real photo from a digitally created one. “I feel like I’m going blind,” he indicated, showing his concern about an AI that is managing to obscure the truth and distort reality. Watermarks as a solution. Faced with this avalanche of images and videos generated by AI that are indistinguishable from reality, one of the potential ways to mitigate the problem continues to gain strength. It is, of course, the watermarkstotally invisible and which are part of the metadata of those files. Two promising initiatives. There are several initiatives in this regard, although the most notable It is that of the C2PA coalition which includes, for example, Google and OpenAI. AI tools should add those watermarks identifying those contents (“This video has been generated with this AI application, this image has been generated or edited with this other one”), but at the moment that type of option is not applied by default. Another important project in this sense is SynthIDGoogle’s technology to “mark” these contents as created with AI. Image | Bild (CC0) In Xataka | What happened to Technicolor: evolution and death of the company that changed cinema and was overwhelmed by its ambition

It is no longer enough to count fingers to know if an image is made with AI. Now you have to learn technical drawing

by

Detecting images generated by artificial intelligence has become a game of cat and mouse. And the worst thing is that it is going to get worse. For a time, we all began to focus on the hands and in the number of fingers that the AI ​​represented in the images of people through the diffusion mechanisms of the models. A few years ago it was obvious to see when an image was created by AI. Now, with image models and video increasingly precise, the task is much more complex. The good news is that there are still ways to detect if an image has been generated by AI, although seeing the pace at which the models advance, this may soon change again. Detecting them is less intuitive than before, but just pay attention to geometry, shadows and perspective. Basically, technical drawing. Who is behind this idea. Hany Farid, a specialist at the University of California at Berkeley and one of the world’s leading experts in image forensics, has spent more than two decades dedicated to determining whether a photo or video has been manipulated. Santiago Lyon, former director of photography for the Associated Press who now works in digital security at Adobe, describes Farid in a Science report as “a kind of dean of digital forensics”, precisely because he has been at it for so long. Farid helped found this discipline more than 20 years ago, and says that AI is the biggest challenge he has faced. Farid exemplifies his method with this image. If we draw a line towards the horizon between the tiles and the skirting boards, we see that the lines do not converge at a single point, which tells us that the image is generated by AI It’s hard to know what’s true and what’s not.. We are losing the ability to trust what we see. The combination of generative AI, capable of creating images almost indistinguishable from reality, and a warm regulation on social networks It makes the hoaxes amplify, making it increasingly difficult to know if what we are seeing is real or not. And in many cases, we don’t even care. Farid speaks directly of a “global war for truth”, with consequences for people, institutions and democracies. In a TED talk He said that he believes that the percentage of fake images on the Internet is close to 50%. It is no longer useful to focus on pixels. One of the first techniques Farid developed was based on the “noise” left by real cameras. An authentic photo is born from light hitting an electronic sensor; An AI image, on the other hand, emerges from a statistical process that converts random noise into an image consistent with the text requested. This very different origin left traces detectable at the pixel level. The problem is that generators have learned to imitate even those imperfections, sensor noise and lens artifacts. As explains Science report, many of Farid’s pioneering methods based on statistical relationships between pixels “no longer work well, if at all,” because AI images are created from scratch rather than edited over a previous photo. technical drawing. AI, says Farid, “doesn’t know physics, doesn’t know geometry, and does all kinds of atrocities.” And that’s where technical drawing comes in. According to Farid, these are the three fronts that we must examine: Vanishing points. In the real world, parallel lines (train tracks, floor tiles, the sides of a wall) converge toward a single point as they move further apart. It is a principle that artists have known for centuries, but that AI ignores because it does not understand three-dimensional space. If those lines don’t meet at a single point, the scene is physically impossible. Shades. The Sun is so far away that its rays reach the Earth practically parallel. That means that the lines connecting each object to the shadow it casts should also intersect at a point consistent with the position of the light. In many AI-generated images, those lines don’t even come close to crossing. Highlights. The same principle applies to mirrors, as lines connecting one point on an object to its reflection should converge at a vanishing point. When they don’t, the image is given away. The same thing happens in this image. If we draw a line that passes through both the vertices of each cube and the vertices of its projected shadow, we see that they do not converge at a single point either. Track accumulation. No technique is infallible on its own, and Farid insist in that the method consists of accumulating clues, as in an investigation. In his TED talk he exemplified this with an image made with AI of several soldiers looking forward. In it he detected the suspicious pattern in the noise, the absence of a coherent vanishing point on the walls and shadows that did not intersect. Three anomalies that gave clues that the image was not real. The underlying reason why this approach stands up better over time is that AI companies are not looking to fool forensic experts like Farid, but rather the average user, since we are at a much lower bar. As he himself says“the visual system forgives all kinds of nonsense in photos because it doesn’t care.” In this image, if we draw a line from a point in the figure to the same point reflected in the mirror, we see that the lines do not converge at a single point either. Doubts and limits. Not everyone in the field shares the same optimism. Some researchers reaffirm that each detection technique has a very short “useful life”, sometimes a few months, because AI improves very quickly. In fact, the famous mistakes on six-fingered hands disappeared in a flash. Farid, however, is skeptical that AI will ever master complex real-world physics, like an explosion, because simulating it is devilishly difficult and companies have little incentive to go that far. Still, he acknowledges that receives a dozen emails every day from journalists … Read more

The Social Security reform has opened the door to working longer. Early retirement will remain half closed

by

Social Security is pushing those who can continue working to delay their retirement as much as possible, but it resists modifying one of the most discussed rules of the system: the penalty in the pension of those who they retire earlyeven when they accumulate more than 40 years listed. The flexible retirement reform contemplated in the Royal Decree 416/2026 will come into force on August 28, launching the Government’s strategy to extend working life of workers and contain pension spending. What changes with the reform. The new flexible retirement regulation seeks to encourage more people to extend your working life as much as possible voluntarily and can make part of their pension compatible with a salary, something that current regulations did not allow. The idea is not to force anyone to continue working beyond the legal retirement age, but rather offer more incentives so that those who can and want to do so, keep working. The person who is already retired, instead of stopping working completely, can do so part-time. In exchange, they will receive a salary for their work and a supplement to the proportional part of the pension. In this way, someone retired can obtain a higher income while still active, and will receive 100% of their pension again when they stop working. That is, if someone retired receive a pension of 1,000 euros, and for working 32 hours a week (80% of a full day) they will pay you a salary of 1,000 euros, your pension will be cut in that proportion, but the sum of salary (1,000 euros) and pension (200 euros) will provide you with greater monthly income. The current regulations force you to choose between working or receiving the pension. Put obstacles to early retirement. The demographic pyramid in Spain, in which there are fewer and fewer young people to maintain the pension system and a longer life expectancy, has forced successive governments to take measures to maximize working life of employees to continue contributing. This has led to the extension of the retirement age, which has been progressively delayed since 2011 to go from 65 to 67 years in 2027. The other measure approved in the pension reform of 2024 to discourage early retirement is to apply some reducing coefficients to the retirement pension, so that the more you anticipate retirement, the less pension you receive in return. Contribute 40 years without reward. One of the problems posed by the application of reducing coefficients is that those workers who already exceed the maximum limit of years of contributions necessary to access ordinary retirement at age 65 (38 years and six months or more by 2027), will not be able to retire early. without penalizing themand end up getting paid a lower pension than other workers with fewer years of contributions. This group has already organized under the association Asjubi40 and different political groups with representation in Congress have carried out proposals to eliminate this grievance to workers with long contribution periods when they want to advance their retirements. As and how he published The Independentvoluntary early retirees bear an average reduction coefficient of 11.36% and receive an average pension of 2,002.58 euros per month, after retiring at an average age of 63 years and two months. In the case of involuntary early retirement, the average reduction rises to 18.9%, the average pension stands at 2,100.42 euros per month and the average retirement age drops to 61 years and ten months. The unaffordable cost of stopping working. The reason given by the Government for not eliminating these reducing coefficients It’s simple: removing those penalties would be expensive. The Executive estimates an additional cost of 3,358 million euros per year for Social Security if the reducing coefficients are eliminated for those who retire early after having contributed 40 years or more. Of that figure, 1,345 million would correspond to voluntary early retirement, and 2,013 million would correspond to those retired involuntarily, that is, those who have been affected by ERE, business closures, force majeure or other cases considered by the General Law of Social Security. Social Security cannot assume it. Although Spain is registering record numbers in terms of number of members. It closed 2025 with a budget deficit of 5.58 billion euros. Once again, we are facing a record to be treated of the smallest deficit of the last 14 years, as as highlighted The Confidential. But it is a deficit, after all. However, the incorporation of contributions such as the Intergenerational Equity Mechanism (MEI), has contributedIn 2026 alone, 1,162.23 million euros will be added to the Social Security Reserve Fund, which reached a total amount of 15,267 million euros last March. In Xataka | From the “Great Resignation” to the “Great Early Retirement”: the labor market loses the experience of those over 55 years of age Image | Pexels (Joaquin Carfagna)

“Slaughterbots” are no longer science fiction in Ukraine. Russians wear masks to avoid the drone that aims at their heads

by

A few years before the start of the war in Ukraine, a Berkeley computer science professor presented at the UN a short called “Slaughterbots”a piece where small drones with facial recognition chased people autonomously. Many saw it then as another technological exaggeration in the style of the Black Mirror series. A few years later the short… has fallen short. Drones that search for tanks, search for people. For much of the Ukrainian war, drones were seen as a support weapon intended to destroy armor, correct artillery fire or monitor enemy movements. That phase has gone disappearing quickly. What is now emerging is something much more disturbing: cheap drones, produced by the millions, designed specifically to hunt down and kill soldiers. individually. They counted in Forbes that the Russian military channels themselves they are warning of Ukrainian FPVs equipped with thermal vision, reconnaissance systems and munitions capable of firing explosive projectiles at a distance directly against a human body. The detail that is generating the most fear is not the weapon itself, but the possibility that these drones are already learning to identify Where to hit to maximize lethality. The idea of ​​small autonomous devices “hunting” specific people no longer belongs to technological dystopias or viral YouTube videos: it is beginning to form part of the front line’s routine. A gigantic aerial hunting area. The most profound consequence of this revolution is that huge parts of the front have been transformed in “kill zones”those corridors where any human movement can be detected and destroyed from the air in a matter of minutes. Ukraine has especially perfected this model around cities like Kostyantynivka or Chasiv Yarwhere small Russian groups are identified long before approaching the defensive lines. The result has been devastating for classical Russian doctrines: large armored columns and mechanized assaults have become too visible and vulnerable. In response, Moscow is trying to create their own “drone racers”infiltrating small teams of operators hiding in basements, destroyed buildings or tree lines to build temporary bubbles of local air dominance. In other words, war is no longer just about controlling the terrain, it is about controlling the sky just a few meters above each soldier’s head. The true technological leap. The most important thing about these new systems is not the size of the explosive charge, but intelligence that begins to guide them. Many Ukrainian FPVs already integrate autonomy modules capable of continuing the attack even when the operator loses signal due to electronic interference. Western companies and civilian developers have created relatively inexpensive kits that turn commercial drones into smart munitions capable of automatically locking on and pursuing targets. Until recently, that autonomy was mainly used against vehicles; now the focus shifts to the infantry. Some models use EFP loadsformed explosive projectiles that do not need to hit directly to penetrate protection and kill the target from a distance. That eliminates many of the defenses improvised measures that had proliferated on the front, from metal nets even the famous Russian “turtle tanks”. The problem for soldiers is that hiding no longer guarantees survival: the drone can continue observing, wait for the exact moment and attack when it detects vulnerability. “Slaughterbots” stopped seeming over the top. We said it at the beginning, in 2017 Professor Stuart Russell launched the short film “Slaughterbots” as a warning about autonomous drones with facial recognition capable of murdering specific people. At the time it seemed like a futuristic hype designed to open ethical debates about military artificial intelligence. Nine years later, the parallels are beginning to be uncomfortable even for those fighting on the ground. Russian soldiers develop countermeasures that seem straight out of a science fiction movie: using masks to confuse recognition systems, throwing helmets as decoys, hiding their heads behind obstacles or remaining completely still to avoid thermal tracking. Obsession reflects a huge psychological change. For centuries, a soldier could attempt to protect himself from enemy fire using cover, armor, or distance. Many fighters now feel that there is a camera constantly watching them from above, capable of deciding when to attack and possibly where to do it to ensure death. The industrial and algorithmic battle. The great Russian fear is that Ukraine will manage to combine mass production, autonomy and precision on an unprecedented scale. kyiv aims to manufacture millions of FPVs a year, and that completely changes the mathematics of combat. Whether a relatively cheap drone can chase soldiers with hit rates close to 80%human wear and tear begins to take on industrial dimensions. That is why Russia is desperately trying to build its own drone racersdeploy interceptors and saturate local airspace before moving larger troops. However, Ukraine maintains an advantage in both quantity and technological sophistication, especially in optics, autonomous navigation and aerial interception. What is being seen in the Donbas is not simply a tactical evolution of drone warfare: it is rather the birth of a new form of combat where thousands of semi-autonomous machines continually compete to detect, pursue and eliminate individual human beings. And the most disturbing thing is that this transformation is just beginning. Image | Defense Ukraine In Xataka | Satellite images reveal how much Russia fears Ukraine’s drones. 7,000 km away they are covering their nuclear missiles In Xataka | Ukraine has resurrected one of the oldest tactics of warfare. And he is isolating Russian cities without the need for soldiers

It no longer goes to the Moon, and that is precisely what makes it more important

by

The design of Artemis III has changed a lot with respect to what was initially conceived. To begin with, initially the objective was to carry out the moon landing. Later, the possibility of not landing on the moon was raised, but testing docking with the landing system in lunar orbit. However, this will no longer be the case. It has been decided test everything in low Earth orbit, since the tests can be done just as well there and costs and risks are reduced. However, this is not the only change that has been announced recently about the mission. They are all advantages. The journey to low Earth orbit is easier, but also offers more opportunities. In Artemis III there will be many vehicles involved, since we will not only have the SLS rocket and the Orion ship as in Artemis II. The HLS landing systems of SpaceX and Blue Origin. To travel to the Moon there are very narrow launch windows, while low Earth orbit offers more launch opportunities for all of these vehicles. If it is necessary to postpone a first date, there would not be a long wait. Why so much propulsion? In Artemis I and II, the rocket that propelled the shipthe SLS, had a cryogenic propulsion upper stage, which used cryogenic fuels to give the rocket an extra boost and escape the Earth’s gravitational pull. But this is no longer necessary in Artemis III. They don’t need to escape from Earth, on the contrary. Therefore, this stage has been eliminated in order to save on fuel. Be careful with the gap. You cannot remove a stage from the rocket and leave it as is. The rocket is manufactured with a specific weight and height profile. If a piece is simply removed, some of that weight is lost and it becomes unbalanced. Therefore, this propulsion stage It will be replaced by a structural spacer. That is, a piece that weighs and measures exactly the same, but has no function. NASA has already made great progress in its construction. Nothing is thrown away here. The upper stage that has been removed will be used on Artemis IV, since there we will travel to the Moon and all the possible momentum is needed. For the third phase it will be much easier and cheaper to fly without it. Orion+HLS. The main objective of Artemis III will be to verify that the Orion spacecraft and the SpaceX and Blue Origin HLS can dock properly. Once this docking occurs, some members of the crew would go from Orion to the HLS, to verify that the transition from one vehicle to another can be made and to carry out the relevant tests. More permanence in space. Artemis III astronauts will stay in space longer than Artemis II astronauts. This will allow them to obtain additional data on Orion’s life support systems under prolonged manned conditions. Slowly and with good writing. The Apollo missions were a leap into the void, literally. It was possible to go to the Moon on several occasions, but the risk was very high. During all these years, a more exhaustive study of the Moon has been done to design the best travel conditions. The goal was to return to our satellite, but not at any price. The return journey has already begun, but it is not worth rushing at the last moment. To do this safely and send many more manned missions in the near future, it is important that Artemis III is a safe test. That’s why these NASA design changes are so important. Of course, we cannot forget that NASA is not the only one responsible. SpaceX and Blue Origin must also demonstrate that their vehicles are safe. Time is running out for everyone. Image | POT In Xataka | We have not yet colonized the Moon and we have already filled it with garbage: there are even abandoned golf balls

memory no longer wants to live in each machine

by

For many of us, memory shortage It may first sound like a problem close to domestic consumption: RAM modules, components and devices conditioned by an increasingly stressed demand. But the phenomenon that The Next Platform describes also points to the other end of the chain. It reaches the large technology companies that train, deploy and offer artificial intelligence models in data centers. The cloud is not an abstraction, and its appetite for memory is forcing us to think about something that until recently seemed unintuitive: perhaps each machine should not depend only on the RAM it has inside. Memory changes places. The underlying idea is to transfer to memory a logic that is already familiar to us with storage. Today, data can live on the computer itself, on another machine on the network, or on a shared system accessed by several servers. The next generation of servers could treat RAM in a similar way: keep a portion local to each machine, but bring a much larger portion to large external systems capable of distributing capacity according to the need of the moment. From there comes what some call “memory godbox”: a large box or cluster of memory that is no longer tied to a single machine. The CXL moment. For years, Compute Express Link has advanced slowly, almost as a promise for more flexible architectures. The technology was introduced several years ago, but current memory pressures are giving it a much more favorable context. CXL provides a coherent interface to communicate processors, memory, accelerators and other peripherals, relying on PCIe. The final idea is simple to tell, although complex to execute: separating resources without breaking the feeling that they work together. CXL didn’t arrive all at once. It was first used to expand the memory of a server using modules connected to compatible PCIe slots. Then, with CXL 2.0, pooling appearedthat is, the possibility of pooling memory in a common pool and assigning it to different machines as needed. The limit was that that memory could be reallocated, but not truly shared between two systems working on the same data. CXL 3.0 It is the point at which that frontier begins to move, because it introduces broader topologies and shared memory between machines, although with certain technical limitations. The underlying problem. According to The Next Platform, AI does not fall short only because of a lack of calculation, but also because of a lack of memory. The HBM that accompanies the GPUs is very fast and is designed to power these chips at high speed, but its capacity is limited and its cost is high. In training, the big challenge is usually processing enormous amounts of data to build the model. In inference, however, we talk about something else: using that already trained model to respond to a request. The memory of the conversation. Each response from a language model is built little by little, token by token. In order not to recalculate everything above at each step, the systems save a type of working memory called KV cache. The Next Platform explains that previous attention vectors are preserved there, which help the model to continue taking into account the context while generating the response. The problem is that in services with many users, this cache can grow to occupy enormous amounts of memory, even more than the model itself. It’s not just theory anymore. This idea no longer lives only in technical documents or architectural promises. The Register mentions Panmnesia, Liqid and UnifabriX as companies working on systems to take memory off the server and make it available to multiple machines. Some do it with CXL switches, others with large reserves of DDR5 that can be distributed among different hosts. The Next Platform adds the case of Enfabrica and its Emfasys system, designed for inference and capable, according to the media, of reaching 18 TB of DDR5 per memory server and 144 TB in a full rack. The conclusion is simple: the industry is not only looking for more memory, it is looking to place it in another way so that AI can take better advantage of it. Images | Xataka with Nano Banana In Xataka | The ‘Chinese Netflix’ has designed a plan for AI to generate the majority of its content within five years. It sounds risky

The mouse cursor has hardly changed for half a century. Google just tried to make that no longer the case

by

Google DeepMind has published the principles and demos of Magic Pointera mouse pointer powered by Gemini who understands what you are pointing out and why. Without writing anything. Just pointing. Why is it important. The chatbot as the main interface has been the dominant model in AI for two years: you open a window, write and you get a response. Magic Pointer proposes the opposite: the AI ​​moves with you around the screen, reads what is in front of you and acts without you explaining the context. If it works as promised, the text box is no longer the gateway to AI. The logic behind the project is that the problem with current AI is not its capacity, but the friction to use it. Every time you want to ask a model for something, you have to drag your world into it: open a window, paste text, explain the context from scratch, etc. Magic Pointer reverses that flow: the AI ​​goes where the cursor is. In detail. The system captures visual and semantic context around the pointer. You indicate a date in an email and Gemini suggests creating an event. You select two images, a sofa and your living room, and the model composes them. You hover over a table and you can request a graph without opening any more apps. The objective is to replace the prompts long by what DeepMind calls “natural shorthand”: point out something, say what you want, and have the system fill in the gaps. There are live demos at Google AI Studio and the system now reaches Chrome. In autumn it will land in Googlebookthe new Google laptop with Acer, Asus, Dell, HP and Lenovo as manufacturers. Between the lines. We are looking at three ways to put AI in a computer: Apple integrates it within each application. Microsoft puts Copilot on a side panel. Google puts Gemini inside the pointing device itself: it is not in the background, it is the cursor, it is the widgetis the interface between the user and the machine. That last one is a philosophical bet. And it has implications for the chatbot model: if the cursor acts as a contextual agent, the chat window loses its monopoly as an entry point. Yes, but. Googlebook arrives in autumn as a premium product, with no announced price yet. The Android ecosystem on the desktop remains the weak flank: if developers do not build native apps for the big screen, the Magic Pointer points to a world that does not yet exist. And in any market where Gemini is restricted by regulations, the entire proposition becomes empty. In Xataka | The AI ​​industry already knows how to make more money. Just use the fear strategy Featured image | Google

Vigo already has its “direct” AVE to Madrid without stops in Castilla y León. Now it takes longer than before

by

A little over a year ago, controversy arose between autonomous communities. Abel Caballero, mayor of Vigo, claimed that one of the reasons why the AVE to Madrid did not take the promised time was due to “an excessive number of stops in the Castilla y León area.” The response was not long in coming. Now, the people of Vigo have their “direct” train to Madrid, without stops in this autonomous community. The problem: they come out earlier and take longer than before. 215 minutes. That is the promise with which the railway link between Vigo and Madrid was announced. Yes, you read correctly: three hours and 35 minutes. Since high speed was launched in this corridor, the trains that have approached the promise that Óscar Puente put on the table in 2024 have been an exception. Instead, the trains have taken at least four hours. In the best of cases, barely a minute has been cut from the journey, but in practice they continue to move on the psychological border of 240 minutes. As long as there is no delay. And in the Galician corridor they know well what we are talking about. Too many stops. Given the impossibility of fulfilling that promise, Abel Caballero, mayor of Vigo, proposed a simple formula: that the trains not stop in Castilla y León. He did not say anything about Galicia because, in his opinion, it was the arrests in the neighboring autonomous community where the most time was wasted. “The current travel time on some of the routes is very long due to an excessive number of stops in the area of ​​Castilla y León, an area already close to Madrid that currently has a very important coverage of trains coming from all over the north and the rest of Galicia on high-speed routes” The words were picked up in the local media Atlantic.net and, according to Caballero, they were well received by Álvaro Fernández Heredia, president of Renfe since where they quickly denied this possibility. The proposal collided head-on with Alfonso Fernández Mañueco, president of Castilla-León, who in words reported by The Spanish He described Caballero’s words as “intolerable.” Vigo-Madrid route starting May 20 “Straight”. So, in quotes. Because starting May 20, Renfe will have an AVE without stops between Vigo and Madrid. The long-awaited Galician demand has been heard. As you see in the image above, from that day on the train from Vigo will leave at 5:50 a.m. and arrive in Madrid at 9:55 a.m. Once it leaves Vigo, the train will stop only in Galician cities. Once we have passed Ourense, the train will not stop until it reaches the Chamartín station in Madrid. It is, finally, the “direct” AVE that the people of Vigo demanded to be able to travel back and forth to Madrid in the shortest time possible. Vigo-Madrid route until May 19 Sure? And, contrary to what logic says, Vigo-Madrid takes longer than before. And, in addition, it will force passengers to get up earlier. And until that day, the first train that connects Vigo with the capital will continue to stop in Zamora but its journey will take eight minutes less. In addition, it leaves at 6:00 a.m., instead of 5:50 a.m. The reason is that, with the reorganization devised by Renfe, the first train has to stop in Santiago de Compostela, which has been omitted until now. And the residents of Ourense do not see a great reduction in times either, since it will only take four minutes less to reach Madrid despite not stopping in the Castilian-Leonese city. To make matters worse, the residents of Pontevedra will not find a substantial advantage either. In fact, they will now take eight minutes longer than before the last change. Now, with the latest changes, all trains leaving in Vigo and arriving in Madrid will take more than four hours. Adif, in the spotlight. Part of these “delays”, they point out from Vigo Lighthouseare due to the works that are being carried out in the Guadarrama tunnel at the entrance to Madrid. These add seven minutes to the final amount but as they point out from The Region There are numerous temporary speed limitations on the Galician section that hinder the promises of connecting Vigo and Madrid in half an hour less. He Vigo Lighthouse He also points out that Renfe has been complaining for some time about Adif’s management of the roads. According to the Galician media, the operator has requested on several occasions that some crossings between trains be relocated so that they are not on the single track sections. And the passage times have not been updated either, so, they point out, the current speed cannot be increased either. An exception. Since the Galician high-speed corridor began to operate at full capacity thanks to the use of the Talgo Avril that can change track gauge, doubts about the reliability and performance of these trains have been on the table. First for its disastrous arrivalthen for their problems with the change of year in 2025 and finally due to the cracks that appeared in these trains on the Madrid-Barcelona. However, the trains have proven capable of operating at full capacity and getting closer to the famous 215 minutes promised between Vigo and Madrid. Last November, A Renfe AVE managed to cover the journey in 217 minutes. The problem is that it only served to alleviate the delay accumulated at the origin. Being the last service and with the tracks already clear along the entire route, the Avril was able to travel at maximum speed for as long as possible. Photo | André Marques In Xataka | Renfe has found a scapegoat for its problems on the Madrid-Barcelona line: Talgo and its AVRIL trains

For the CEO of Ford, the reference for the electric car is no longer Tesla, it is China

by

The head of Ford has been studying Chinese manufacturers in depth for months and is clear about one thing: that to understand where the electric car is going, we must pay close attention to China. For some years now the country is leading a historic transition in the automobile, and the perfect proof of this reality is the fixation that brands as historic as Ford have with the Chinese electric car. And for Jim Farley, CEO of the company, Tesla is no longer the benchmark. China, not Tesla. The automobile industry has been at a crossroads for some time. Electric sales are not growing at the expected rate in the West, large manufacturers have had to rethink their strategies and convert their factories (energy storage for data centers), and in the United States the elimination of federal tax incentive It has made the purchase of a new electric car even more expensive. In this context, Ford CEO Jim Farley explained in the Rapid Response podcast that Tesla is no longer the benchmark, and that it is now China. Change of sight. In the interview, Farley explained why he has been testing a Xiaomi SU7 instead of an American vehicle. “If you’re an American and you want us to beat the Chinese in the car business, you’re going to want to pay attention, not necessarily to Tesla. Nothing against Tesla, they’re doing well, but they don’t really have an up-to-date vehicle,” he said. And his reference for Ford is not Elon Musk, but BYD: “The best thing in the business for us in cost, supply chain, manufacturing experience and innovation is BYD,” Farley said. in the same podcast. Concerning. BYD was born in 1995 as a battery manufacturer and today is the largest electric car manufacturer in the world by volume. having surpassed Tesla in global sales in 2025. In 2022 it was the first manufacturer to completely abandon pure gasoline cars. For Farley, what is relevant is not the market capitalization of each company, but rather who is defining what the consumer will want to buy in the next decade. TOGod to the expensive electric ones. Ford has learned its lesson through million-dollar losses. The company became the second brand that sold the most electric cars in the US after Tesla, but its models were, according to Farley himself, “designed in the wrong way.” In December 2025, Ford took over a $19.5 billion correction having to reformulate its entire electric strategy. He F-150 Lightningwhich was presented as the flagship of its electrical commitment, is converted into an EREV vehicle (with a small combustion engine that acts as a generator) because, as admitted Farley himself in December, “the $70,000 electric cars were not selling.” The new roadmap involves launching an electric pickup at $30,000 before 2027. The key is in the second-hand market. Farley has an unconventional way of reading the market. And it is that prefer look at the sales of used cars before those of new ones, because “the second-hand market is twice that of new ones, and since they are all sold at lower prices, they are a better predictor of consumer behavior.” And of course, in this market, affordable electric and hybrid vehicles are the ones that move the most compared to those in the premium segment. China is not just price. Farley recognize that each Chinese car incorporates about 4,000 or 5,000 dollars in government subsidies, direct and indirect. He is also aware that these vehicles incorporate up to ten cameras and advanced connectivity systems that, in his opinion, “should be reviewed by the US Department of Defense for reasons of national security.” However, Farley concludes that the correct response is not to ignore them, but to learn from them. “That is the gift that China has given us: that we are respectful enough of its progress not to settle for business as usual,” he said in the interview. Cover image | Hans and Rapid Response In Xataka | The longest straight road in the world is a mental challenge: 240 km without curves, in the middle of the desert and with truck traffic

Taxis have always had four seats. With robotaxis that no longer makes sense.

by

When Tesla taught the Cybercabthere was a detail that was obvious even before talking about autonomy, sensors or commercial deployment: I only had two seats. It was not a minor decision. For decades we have associated the taxi with a car capable of carrying four passengers, with its driver in front and a back seat designed for almost everything. That’s why that model without a steering wheel or pedals seemed, at the very least, a rarity. Now, seeing what we have seen later, that image begins to have another reading. The interesting thing is not only that we are talking about driverless cars, but about vehicles that can be thought of differently from the first moment. The traditional taxi took advantage of a well-known architecture: four or five seats, a driving position and a body prepared for very varied uses. On the other hand, a robotaxi designed for a fleet can ask a more precise and specific question: what do most journeys require? The answer appears when we look at how these services are used. Some time ago, Lucid’s Marc Winterhoff and Uber’s Andrew Macdonald, they pointed out that more than 90% of the journeys Uber offers only one or two passengers. This proportion helps to understand why two-seater robotaxis are beginning to appear in more projects. It is not advisable to turn this information into a universal rule, because each city, service and use case has its nuances. In that context, Tesla’s proposal fits better. The Cybercab is not planned as a conventional car to which autonomous driving is simply added, but as a vehicle designed to operate, if Tesla manages to deploy it as promised, within a transportation network without a human driver. Hence it dispenses with a steering wheel and pedals, and reduces the cabin to two occupants. The firm led by Elon Musk presents it as a specific piece for safe point-to-point travel. Tesla is not the only one who has reached a similar conclusion. Lucid showed in March 2026 Lunara two-seater robotaxi concept without steering wheel or pedals, although it should be stressed that we are talking about a conceptual proposal and not a product ready to hit the streets. Verne, the Croatian company linked to Mate Rimac, also previously presented a two-seater electric autonomous vehicle. Lucid Lunar The logic behind these designs is not only spatial, it is also economic. A smaller vehicle may require fewer materials, move less weight and consume less energy per trip, something especially relevant when we talk about fleets that should circulate many hours a day. Lucid, for example, presents Lunar as a vehicle designed to be as efficient and cheap as possible for fleet operators. The company projects an efficiency of 8.9–9.7 km per kWh in typical use, although these are figures from a conceptual proposal and not from an already deployed fleet. Furthermore, the change is not limited to the size of each vehicle. It also affects the way we imagine fleets. A study by Boesch, Ciari and Axhausenlinked to ETH Zurich, modeled a specific scenario in Zurich and concluded that, if waiting times of up to 10 minutes were accepted and adoption was sufficiently wide, a fleet of shared autonomous vehicles could very significantly reduce the total number of cars needed, even up to 90% in certain conditions. It is not a universal recipe, but it is an important clue: the robotaxi not only rethinks the seat, but also the scale of the system. So we could say that the four-seater taxi will continue to make sense for many uses, and the fleets of the future will probably need to combine different vehicles. The novelty is that the robotaxi allows each need to be better separated. For individual or two-person trips, a smaller model may be sufficient, more efficient and easier to justify within an on-demand network. What a few years ago seemed like a strange decision begins to fit with another way of looking at mobility: not always designing for the maximum possible, but for what happens most of the time. Images | JavyGo | Maxim In Xataka | Xiaomi CEO has a message for the world: the “cheap electric car” may never arrive

Log In

Forgot password?

Forgot password?

Enter your account data and we will send you a link to reset your password.

Your password reset link appears to be invalid or expired.

Log in

Privacy Policy

Add to Collection

No Collections

Here you'll find all collections you've created before.