months

The new Galaxy A debut with discounts of up to 345 euros and financing for up to 24 months without interest

by

The new ones Galaxy A57 and Galaxy A37 They are now available in stores. Normally, buying a mobile phone or any output device means doing so at its RRP or at a price very close to it. However, in recent times we are seeing very interesting launch promotions, the kind that come in handy to get us something brand new with very good conditions. For example, you can take a Galaxy A37 256GB with a discount of up to 245 euros or a Galaxy A57 512GB with which you can save up to 395 euros thanks to Delivery and Premiere. The price could vary. We earn commission from these links And be careful, because if that doesn’t seem enough to you, you can also finance for 24 months without interest. We tell you more about this promo. The Galaxy A57 is Samsung’s best mid-range He Galaxy A57 has come to be the spearhead of Samsung’s mid-range. Curiously, It is the one with the most discounts so that we can get it. This device starts, with the launch promo, at a price of 529 euros for its cheapest version, which has 128 GB of storage and 8 GB of RAM. Below, so that you can see it more clearly, we leave you the prices that the three versions have right now: Galaxy A57 (8+128GB): 529 euros. Galaxy A57 (8+256GB): 559 euros. Its RRP is 589 euros. Galaxy A57 (12+512GB): 679 euros. Its RRP is 769 euros. With the prices on the table, let’s see the discounts that we can apply using the model with 512 GB of storage as an example: Since its RRP is 769 euros, we will already be saving 90 euros right from the start. If we use Delivery and Premierethe renewal service that the Samsung store has, we will get a direct discount of 50 euros. This is independent of the refund that we will receive for our old mobile phone, which may be up to 155 euros. If we use PayPal or Bizum as a payment method, we will obtain another additional discount of 50 euros. In addition, we also have the code ‘SAMSUNG5’ available, with which we will receive 5% discount. Adding all this together, the result we get is that we can get the Galaxy A57 with more memory for 550 euros (or 22.92 euros per month if we finance for 24 months with PayPal). A great price to which, we repeat, we can add the refund we receive for the mobile phone we deliver. The price could vary. We earn commission from these links The Galaxy A37, from 360 euros Now it’s time to talk about the Galaxy A37, a more economical option than the previous one. In this case, we are dealing with a mobile phone that has two configurationsboth with 8 GB of RAM and different storage. The launch prices from which both start are as follows: We are going to place the discounts below, which vary a little from those of the Galaxy A57: If we choose the version with more memory, we will already be saving 45 euros from the beginning. With Delivery and Release, we will receive a direct discount of 25 euros. Furthermore, the maximum refund that we can receive for our old mobile phone is 150 euros. Using PayPal or Bizum to pay, 25 euros discount. In addition, we can also use the code ‘SAMSUNG5’ to receive a 5% discount. With all of the above, the price will remain more than attractive. 403 euros or 16.79 euros per month if we finance for 24 months. All without forgetting the valuation of the device that we deliver, which in this case may be up to 150 euros that we will receive in the form of a refund. The price could vary. We earn commission from these links Some of the links in this article are affiliated and may provide a benefit to Xataka. In case of non-availability, offers may vary. Images | Samsung In Xataka | Best wireless headphones. Which one to buy and 21 models from 15 euros to 470 euros In Xataka | The best mobile phones, we have tested them and here are their analyzes

The CNMV has tested AI to invest in the stock market for ten months. The conclusions are very revealing

by

In recent months there has been a recurring discourse that we see on social networks and that sell us again that “get rich quick” message. That message is “use AI to invest in the stock market.” The interesting thing comes when we see how the CNMV has published a study in which it has precisely attempted to analyze that premise. Although this organization warns of the risks of investing with AI, there is another important message in the conclusions: LLMs are not bad investors per se. They are bad at following vague instructions, which is just how most people use them. The CNMV study. Two researchers from the CNMV, Ricardo Crisóstomo and Diana Mykhalyuk, have published a study methodologically serious (but imperfect) and very interesting: they used four AI models for ten months live, from April 2025 to January 2026. They chose ChatGPT, Gemini, DeepSeek and Perplexity as models. The process was simple but demanding: each month they asked each model to identify the five stocks in the Ibex35 index with the best expected performance (to buy) and the five with the worst expected performance (to sell short). Then the real result was measured at the end of the month, and here there was no historical data selected just because: the real market was the only arbiter of all the functioning of the models. The models evolved. One of the most significant aspects of the study is that its creators recognized a methodological problem that was difficult to avoid: during those ten months, the versions of the four models were updated several times. The Gemini of April 2025 was not the same than that of January 2026for example, and that could influence the results. The researchers commented that it was impossible to know with certainty whether an improvement or deterioration in performance was due to the prompt strategy, market conditions in that period, or simply because the model changed. The prompt is everything. Three were also tested prompt types very different, and that gave rise to conclusions that were neither alarmist nor did they create false expectations: they were “it depends.” Thus, their results showed that everything depended on the type of supervision that these models had: If the LLMs were asked generic questions such as “What stocks should I buy?”, they failed repeatedly. There were computational errors, incorrect interpretations and also the famous hallucinations of chatbots. Curiously, the only one that made a profit was ChatGPT. The problem is that people who use AI to invest probably use this mode of action. But if prompts prepared with iterative reviews and human supervision at each step were used, Perplexity achieved a monthly return of 3.5% on the IBEX35. Gemini and ChatGPT also improved their behavior if given more precise instructions, and DeepSeek was the worst ranked overall. There is another finding: when models receive official regulatory documentation or business results reports, their predictive accuracy improves significantly. The LLMs they reason better on concrete and verified facts than generating analysis from scratch on information that they themselves search for on the web. financial hallucinations. The CNMV study points out that financial markets are especially demanding for AI models because they require complex processes. They have to retrieve and collect information dynamically, they have to reason in multiple steps, they have to be numerically precise, and they have to know this market, and all in real time. Chatbots are trained to generate “convincing” textsso the incentive here is that the investment recommendation “sounds good” even though it is completely wrong. The confidence with which AI models present incorrect financial analysis is proportional to the risk they pose to those who use them without checking whether what they say makes sense. In short: do not trust AI to invest right off the bat. The Reddit user’s experiment was equally striking, but hardly conclusive. Source: Reddit. The Reddit experiment. A Reddit user named Blotter-fyi rode in November from 2024 a platform called Rallies.ai which gave several AI agents access to real-time financial data and money to make stock market operations. Four months later, with the S&P index down 7% since the start, five of the models are outperforming that index, although only two have positive returns in absolute terms. The author himself was the first to warn that four months are insufficient to reach a conclusion: it could be luck, the market or simply the prompt. Nof1’s experiment was fascinating, but it made it clear that AI models don’t typically make money investing in crypto. Source: Nof1. Nof1 and crypto fascination. Another particularly striking experiment was the one that the company nof1.ai made with its Alpha Arena. He put six AI models to compete, gave them 10,000 real dollars each and gave them two weeks to trade cryptocurrency derivatives without human intervention. The most striking result was not who won, but who lost: GPT-5 ended with more than 25% losses and Gemini with close to a negative 40%. Meanwhile, the Chinese models Qwen and DeepSeek dominated in terms of good performance. They iterated with other models, 32 in total, and of all of them only six achieved a positive return: the rest lost money. Grok-4.20 was the big winner ahead of GPT-5.1 and DeepSeek v3.1. Maybe you shouldn’t just let AI invest for you. The conclusions after these experiments are clear. Four months of a model outperforming the S&P index in a bear market does not prove that AI is a good investor. Only in that specific period, with that specific marketthat model made decisions that turned out to be less bad than those in the index. To see if this makes sense takes years, multiple market conditions, and many instances of the same experiment running in parallel. The same happens with Nof1 – especially short – and with a more serious and methodical process like that of the CNMV, which was also surrounded by events whose impact on the final result was uncertain. Faced with so many unknowns, the conclusion seems clear: … Read more

In three months of 2026 he has earned more than in all of 2025

by

There are brands that are part of our routine almost without us realizing it. Samsung is one of them. For many, it is the cell phone we carry in our pocket, the television in the living room or that appliance we use every day. Therefore, when we look at what is happening with the company in the first months of this year, the surprise is inevitable. In a context where a good part of the technology industry deals with rising costs and certain instability, Samsung is projecting results that have significantly exceeded forecasts. And this is where the story is truly understood. This mismatch with the more restrained tone that a good part of the sector carries is not coincidental, and the data helps put it in context. We are not just talking about an impression, but something that is clearly seen. As soon as we stop to look at the numbers calmly, the picture changes and what is happening with Samsung these days begins to make sense. The rise of artificial intelligence has clear winners and Samsung is one of them The South Korean giant estimates that its operating profit in the first quarter of 2026 could be around 57.2 trillion won, about 37.9 billion dollars, compared to 6.69 trillion won (4.525 million dollars) recorded in the same period of the previous year. The figure exceeds the 43.6 trillion won that the company obtained in all of 2025, which implies that In just three months he has earned more than in the entire previous year. In parallel, revenue would also advance strongly with growth of nearly 70% year-on-year, and above 100 trillion won in a single quarter for the first time. The impressive jump in Samsung’s profit in 2026 after several years of ups and downs | Graphic: Xataka | Source: Samsung/Blooomberg It is important to understand well what we are talking about. Operating profit measures how much a company earns from its core business, before taxes, interest and other financial factors. That is, it gives us a pretty clean idea of ​​how the business itself is working. It is not the same as the net profit, which does include all those adjustments and is the final figure. In the case of Samsung, these data are still preliminary: the company will publish its complete results, with the breakdown by divisions, on April 30. But it’s not enough to look at the accounts, you also have to look at the business. Samsung not only sells devices, it is also one of the largest memory manufacturers in the world, an essential piece in any technological infrastructure. And this is where the story changes scale: a good part of that memory does not end up in cell phones or televisionsbut on servers and data centers that support AI services. It is a business that is less visible to the general public, but much more decisive at this time. What we are seeing, in reality, is the direct impact of that other Samsung, the one that operates at the base of the current technological revolution. The key is to understand that production capacity is limited. As Micron explained a few months agomanufacturers cannot multiply their production from one day to the next, so they have to prioritize. And right now a good part of the industry is directing its resources towards AI. The systems that make it possible need large amounts of advanced memory, especially HBM, and that has pushed manufacturers to focus on that segment. It is not only a technical issue, but also an economic one, because these chips offer better margins and much more intense demand. The side effect appears immediately. If an increasing part of the capacity is dedicated to that advanced memory, other products take a backseat and supply becomes strained. That is exactly what is happening with DRAM, one of the most widespread types of memory in consumer electronics. According to Citigroup, quoted by Bloomberg, its global average price rose by 64% in the first quarter compared to the previous one. The consequence is direct: manufacturing mobile phones, computers and other equipment becomes more expensivewhich puts pressure on margins and forces us to review costs, configurations or prices. It is not worth losing sight of the fact that Samsung is a South Korean company, and that is more important than it seems. We are talking about the largest company in the country and one of the best reflections of the technological muscle that South Korea has built around semiconductors. In addition, it does not play alone: ​​it competes in the same league as other large memory manufacturers such as SK Hynix, also South Korean, and Micron, in the United States. A good part of the memory used by the world is shared between these actors, which turns their decisions into something that goes far beyond their own accounts. If we think about it for a moment, it makes a lot of sense. All of this AI fever is being built on top of data centers filled with very specific hardware. NVIDIA is the clearest example, because its chips are at the center of that infrastructure and have captured much of the attention. But those systems don’t work alone. In order to train models and operating on a large scale require enormous amounts of memory, and that’s where Samsung fits. It does not occupy the symbolic place that NVIDIA has today, but it does benefit from the same wave of investment from a less visible and, as we have seen, very profitable position. Images | Xataka with Bano Bana | Samsung In Xataka | Europe cannot be a “technological vassal of the United States”, and the CEO of Mistral is clear about the path

Microsoft’s problem is not having lost a quarter of its value in three months. It’s just that he’s been wrong for a long time.

by

It seems like not so long ago when many celebrated Microsoft’s commitment to Azure. The decision of Satya Nadella Focusing on cloud computing soon began to translate into good financial results, propelling the Redmond company to achieve record revenue figures. But there was something more relevant in that movement: the realization that it could generate enormous benefits beyond Windows. That strategy, started in 2014ended up marking a before and after that became especially visible in 2019, when the firm reached for the first time a market capitalization of one trillion dollars. However, not even the most long-term oriented strategists, like Nadella, are free from errors. Microsoft has been chaining questionable decisions for some time that have ended up having a direct impact on its quarterly results. Specifically, the company has lost almost a quarter of its value in just three months. To put it in context, we are talking about its largest quarterly drop since the 2008 financial crisis. A decline of this magnitude, logically, does not go unnoticed. From cloud leadership to a strategy under pressure If we want to understand why the story has gone wrong, we have to start with the most obvious: the market has reacted harshly and, above all, selectively. In the first quarter of 2026, Microsoft lost about 23% of its stock market value, according to CNBCwhile the Nasdaq lost around 7%. It is not a minor movement, among other things because we are talking about a drop of a magnitude that has not been seen in almost two decades. This gap compared to the rest of the sector begins to point out problems that go beyond the general context. For a time, the commitment to OpenAI was seen as one of Microsoft’s great strategic successes, and it is not difficult to understand why. The company has invested around 13 billion dollarss to integrate this technology into Azure and into products like Copilot, which allowed it to place itself in a very advantageous position in the race of the artificial intelligence. However, with the passage of time we have also begun to see the other side of that decision: a very high technological dependence and a growing pressure to justify that deployment. As the months have passed, that close relationship has also quietly begun to change. Although Azure remains a key partner for OpenAI, the company led by Sam Altman has started to open your infrastructure to other actors to sustain the growth of its models, which increasingly require more computing capacity and energy. This does not break the alliance, but it does change its meaning, because Microsoft no longer concentrates with the same clarity all the strategic advantage that it had achieved in the first phases of the agreement. If we go down to the field of the product, where all these bets should materialize, the case of Copilot is especially illustrative. Microsoft has tried to make this assistant the axis of its new value propositionintegrating it into Microsoft 365 and a good part of its ecosystem, but the adoption It is not going at the expected pace. According to The Information, almost no one uses Copilot. What we have seen is that bringing artificial intelligence to the daily life of companies is more complex than it seemed on paper. Added to all this is a tension that is not always seen, but is very present in the backroom of this race: that of how to distribute resources in an environment of growing demand. Microsoft is investing massively in infrastructure to sustain the rise of AI, but at the same time it has to decide how it allocates that capacity between Azure and its own services. In January, CFO Amy Hood came to point out that Azure’s growth in the December quarter would have been even greater if the company had allocated more chips to the cloud instead of distributing some of that capacity among services like Copilot. Attrition is not limited to artificial intelligence, and that should also be taken into account. Also this year we have seen notable drops in income and in various areas of the Xbox ecosystemin a context also marked by previous price increases in Game Pass and on the consoles. It may seem like a minor front next to Azure or Microsoft 365, but it helps complete the picture of a company that has been opening too many flanks at the same time. What we have seen is that even in areas where it had a consolidated position, Microsoft is finding it more difficult to keep pace. Put all these pieces together, and what begins to emerge is an increasingly evident disconnect between Microsoft’s operational strength and the way the market is valuing its strategy. The company remains the fourth most valuable on the planetcontinues to grow, with revenue up close to 17% year-on-year in its last reported quarter and with Azure advancing 39% in the December quarter, but that strength is not translating to its price or valuation. Images | Xataka with Nano Banana 2 In Xataka | The ghost of IBM: Satya Nadella’s great challenge is to prevent Microsoft from becoming a technological fossil

Three months ago Australia banned social media for those under 16 years of age. It is already investigating possible breaches

by

Just three months ago, Australia launched one of the most ambitious regulations that have been proposed so far on social networks and minors. The measure came into force on December 10, 2025 with a clear message: force platforms to prevent those under 16 years of age from having accounts and give families back part of the control over the digital lives of the youngest. From the first moment it was presented as a pioneering initiative, but something important was also assumed from the beginning: applying it was not going to be easy. The first doubts. The rule has already entered its most delicate phase, checking whether it is really being applied as planned. The eSafety regulator has opened the first formal review and has put platforms such as Facebook, Instagram, Snapchat, TikTok and YouTube under scrutiny. The agency speaks of “significant concerns” and points to failures in control mechanisms. It also points out that current systems are not effectively preventing those below that threshold from continuing to open new accounts. How minors are sneaking in. The report goes beyond a general warning and focuses on very specific failures in the control systems. It has been detected that there are not enough safeguards to prevent users under the permitted age from creating new accounts, but also something more striking: some platforms allow the verification processes to be repeated until the user manages to pass them. Also in certain cases, these profiles are invited to demonstrate that they meet the age requirement even after having indicated that they do not, which shows inconsistencies in the application of controls. A problem that was already anticipated. The difficulties in applying the rule have not arisen now, they were already on the table from day one. When the law came into force, The Australian Government itself admitted that its implementation would not be perfect, and the first signs pointed in that direction. According to ABC, Some minors managed to bypass the verification systems with basic tricks, such as altering their appearance in facial controls. The outlet itself also warned that parents and older siblings could help some children get around the restrictions, an early sign that the challenge was not just in passing the law, but in making it really work. What is at stake for the platformss. The investigation opened by eSafety does not remain a diagnosis, it opens the door to possible sanctions if it is demonstrated that companies have not taken reasonable measures to prevent minors affected by the rule from having an account. Reuters points out that The fines can reach 49.5 million Australian dollars and affect the aforementioned services and platforms. The regulator has already begun collecting evidence and hopes to close at least part of its investigations by mid-year, which places technology companies in a scenario in which non-compliance is no longer just a reputational risk. The Spanish mirror. What is happening in Australia helps to put into context a debate that has also gained weight in Spain, although here it is at a different point. Peter Sánchez announced in February that The Government wants to prohibit access to social networks for minors under 16 years of age within a broader package of measures on age verification, traceability of hate and responsibility of technology managers. The key difference is that that ban has not come into force and is not being enforced. Still, the Australian case offers a useful reference to anticipate what kind of challenges may appear when such a measure moves from political announcement to actual implementation. Images | cottonbro studio In Xataka | “What the hell is happening with Lidl Spain?”: Germans are speechless at the chain’s comic surrealism

Predicting a drought six months in advance was a utopia. The UPV has achieved this with a system that uses AI

by

In recent years drought episodes have intensified in some regions and fear of a global drought flies over the environment. In this scenario, a team of researchers from the Polytechnic University of Valencia have created a system that can predict whether there will be a drought six months in advance. The system. The work has been carried out by the team from the Institute of Water and Environmental Engineering (IIAMA) of the UPV and has been published in the journal Earth Systems and Environment. The method integrates predictions from four reference climate systems (ECMWF-SEAS5, Météo-France System8, DWD-GCF2.1 and CMCC-SPSv3.5) and are processed using artificial intelligence techniques. From this data, the team calculated two of the most important international drought indices (the Standardized Precipitation Index and the Standardized Precipitation-Evapotranspiration Index), using data windows of 6, 12, 18 and 24 months. The method has been applied in the Júcar River basin, which usually goes through stages of recurrent and quite intense droughts. Why is it important. The novelty of this system is that it is not limited to using a single climate model or index, but rather it merges three pieces that are usually used separately and adds AI processing to correct biases and adapt the models to a regional scale. This allows the prediction to be more reliable since it does not depend on a single model. Furthermore, all of this has been integrated into an operational web toolintended to be used in water management and not only as an academic exercise. Results. The system is correct with a reliability of 90% when the prediction is made for that same month. If they want to obtain predictions three months in advance, the reliability is 60%, while for longer periods (12, 18 and 24 months) they do not give a percentage, but they affirm that the model is still useful for predicting what will happen up to six months in advance. Héctor Macián, co-author of the study, states that “The results confirm that the system is especially effective in reinforcing early warning of droughts, a fundamental aspect to anticipate management measures, reduce socioeconomic impacts and increase resilience to climate change.” Action window. As we said, the methodology has been developed in the Júcar river basin, which is a semi-arid area with long, dry and very hot summers, although researchers highlight that it is transferable to other drought-prone areas. Being able to foresee these episodes with up to six months of margin opens a window to implement the drought management plans much more in advance and thus be able to mitigate the effects. Image | UPV In Xataka | The remains of an ancient Mayan city leave us lessons for the future: an amazing system against drought

We have been avoiding the definitive energy crisis for months. Iran’s missile at Qatar’s largest gas plant threatens to detonate it

by

We had been holding our breath for weeks, assuming the logistical tension in the Strait of Hormuz like the new normal. However, the war has crossed an irreversible red line. We have gone from a trade blockade to the physical destruction of the world’s energy engine, and the consequences are already being felt in the global economy. The impact has been immediate. The price of natural gas in Europe (the TTF reference contract) has shot up 35% in a matter of hours, resurrecting the worst ghosts of the Ukrainian crisis of 2022. The magnitude of the disaster is such that Susan Sakmar, a professor at the University of Houston, warns in Bloomberg that this attack could be “a turning point for the LNG sector, similar to the attack against Nord Stream or perhaps even worse”, as it is a sudden interruption with no signs of a short-term solution. The chronological climb. To understand how we got here we have to look at the chain of events of the last 48 hours. The original trigger, as revealed The Wall Street Journalwas an attack by Israel against the South Pars field, the jewel in the crown of the Iranian energy industry, with the aim of suffocating the sources of financing for the Revolutionary Guard. And it is not just any objective. The analyst Joaquín Coronado emphasizes that South Paris (shared with Qatar, where it is called North Dome) is the largest natural gas field in the world, hosting 10% of global reserves. 70% of Iranian domestic consumption gas comes from there and generates 80% of the Qatari State’s income. A withering response from Tehran. As pointed out Financial TimesIran launched ballistic missiles against the giant Ras Laffan industrial complex in Qatar, the largest liquefied natural gas (LNG) facility in the world and home to key infrastructure such as Shell’s Pearl GTL plant. State-owned company QatarEnergy confirmed “extensive damage” and fires at its facilities. Panic spread throughout the Persian Gulf. According to Reutersthe Iranian Revolutionary Guard issued public evacuation orders, declaring vital energy facilities in Saudi Arabia (such as the Samref refinery and the Jubail complex), the United Arab Emirates (the Al Hosn gas field) and Qatar as “legitimate targets.” Shortly afterward, Riyadh intercepted missiles aimed at the Saudi capital. The market has felt the blow. Oil prices have gone crazy. As detailed oil price, a barrel of Brent surpassing the barrier of 110-113 dollars, which represents an increase of almost 60% in this month of March. However, the real problem goes beyond the daily price. Martin Senior, of Argus Media, warns of a “new level of impact”. It is no longer just about the logistical closure of the Strait of Hormuz (through which 20% of the world’s oil passes); The problem is that the time to repair these destroyed facilities could last much longer than the war itself. And the worst omens already have figures. As has revealed exclusively in Reuters CEO of QatarEnergy, the Iranian attack has knocked out 17% of the country’s LNG capacity for a period that could last up to five years. The domino effect. This situation is taking third countries on their way. As explained CrownedIraq has suddenly lost 3,100 megawatts of electricity due to the Iranian supply cut, while Türkiye will be forced to compete fiercely for emergency LNG shipments. In Europe, the panic is evident: the bulletin Europe Express of the Financial Times reveals that war has blown up the EU leaders’ summit in Brussels, where debate on how to improve competitiveness has been completely overshadowed by fear of energy bills and domestic pressure on the emissions trading system. Geopolitics to the limit. Diplomacy appears broken and America’s allies are losing patience. According to the Wall Street JournalArab governments are “furious” because they feel that the US and Israel strategy has put a target on their backs. For its part, Al Jazeera includes the statements of the Saudi Foreign MinisterPrince Faisal bin Farhan, who has warned Iran that the Gulf’s patience “is not unlimited” and they reserve the right to take military action. Qatar, for its part, has expelled the Iranian diplomats, giving them 24 hours to leave the country. In the midst of this chaos, Washington’s role is erratic. President Donald Trump went to social media to deny prior knowledge of the Israeli attack on South Paris. However, how to collect WSJ, Trump issued an ultimatum to Tehran: if it attacks Qatar again, the US will “massively blow up the entire” Iranian oilfield. Faced with rising prices, the White House is seeking desperate measures. The column of Javier Blas in Bloomberg reveals a controversial plan of the US Treasury: to intervene directly in the financial markets by betting on the downside (shorting) in oil futures to artificially make gasoline cheaper before the elections. An idea that experts such as the CEO of CME Group describe as a “biblical disaster” that would destroy confidence in the free market. The peripheral context. To get the full picture, you have to look beyond the explosions. Verisk Maplecroft Analyst warn in Reuters that the greatest danger right now is that the attacks will extend to Saudi Arabia’s East-West pipeline or to Red Sea ports. These were the only viable alternative routes to avoid the blockade of the Strait of Hormuz, through which 20% of the world’s oil normally transits. In an attempt to cushion the blow domestically, the Trump administration has temporarily suspended the century-old Jones Act (Jones Act) for 60 days, allowing foreign-flagged ships to transport oil and gas between US ports to reduce costs. The dead end. The panorama is bleak. As they reflect on Five Daysthe apparent lightness with which this conflict has developed has dragged us into a dead end. Iran has shown that it does not need to win a conventional war; It is enough for him to set the energetic heart of the planet on fire. Even if a ceasefire were signed tomorrow and ships sailed freely through the Strait of … Read more

A woman spent six months in prison because an AI made a mistake. The terrible thing is that no one checked it

by

Angela Lipps is a resident of Tennessee (USA) who has never been on a plane or taken a trip to other states in the country. Even so ended up in a security cell 2,000 km from her home for a terrifying reason: AI facial recognition software decided that her face matched that of a scammer operating in North Dakota. we have it. It all started with the clue given by the security cameras. Fargo police were investigating a bank fraud in which a woman used fake military IDs to withdraw huge amounts of money. The detectives in charge of the case decided to entrust the work of recognizing the images from the security cameras to a AI facial recognition softwareand after the analysis the system returned a name: Angela Lipps. An agent took a look at her social media, decided that her body and hairstyle matched those of the suspect, and signed the arrest warrant. like in the movies. A US Marshals team showed up at Lipps’ home in Tennessee and He detained her at gunpoint. She was babysitting four children, but that didn’t matter: she was treated like a fugitive from justice. They did not ask him any prior questions, nor did they compare his version. They didn’t even have physical evidence that placed her in North Dakota beyond what the facial recognition system had said. And since the AI ​​said it, it had to be true, right? Six months in prison (and in limbo). Being considered a fugitive, Lipps was not eligible for bail, and spent 108 days in a Tennessee jail waiting to be extradited to a state she had never visited. Then, in late October, she was transferred to a prison in North Dakota. In all that time, no one at the Fargo Police Department bothered to even check to see if the suspect had an alibi because once again, there was no need to check: AI couldn’t fail. At least, according to the police forces that were in charge of the case. Zasca. The funny thing is that proving Lipps’ innocence was really easy. When a public defender finally reviewed the suspect’s bank statements, the case fell apart. While the scammer was stealing thousands of dollars in North Dakota, Angela Lipps was buying tobacco at a Tennessee gas station, using Uber Eats and cashing her Social Security check in her hometown. The GPS and bank records were definitive and irrefutable evidence. Come home come back. On December 24, on Christmas Eve, the prosecutor’s office in charge of handling the case dropped the charges and Angela Lipps was released. Of course: they did it without further ado, on the street, in a state she didn’t know and where it was very cold while she had been arrested wearing summer clothes. The defense lawyers were in charge of paying for a hotel for him and another NGO called F5 Project helped him return home. The tragedy does not end there. The problem is that his return to Tennessee was not happy at all. During the six months he spent in prison, Lipps was unable to pay his bills and ended up losing his house, his car, his savings and even his dog. The Fargo police chief, who held a news conference to mark his retirement, did not even want to answer questions about the case. There has been no official apology or compensation for this huge police error. It’s not the first time, but it seems incredible that it won’t be the last.. We do not know what will end up happening with this scandal, but it is not the first of its kind that has occurred. In the US they have been arresting suspects using facial recognition systems for some time, but These systems fail and cause arrests of innocent people. This type of problem of poor application of AI tools in criminal investigations is present in Spain, where we already talked about the tragic consequences of using VioGén or what happened to him false complaint detection system just a year ago. AI can help, but in these types of processes human supervision is especially crucial. Image | Xataka with Freepik In Xataka | AI videos have broken Instagram and TikTok algorithms. Welcome to the new “AI dumps”

Meta spent a fortune on AI talent and data centers. Nine months later the result is: zero models

by

Mark Zuckerberg wanted to be the Florentino Pérez of AI. last summer began to sign galacticos in this segment and getting talent by letting go stacks of millions of dollars. He more popularOf course, it was the AI wunderkind Alexandr Wangwho became leader of its “Superintelligence” division. The funny thing is that the months go by and go by and in Meta they don’t seem to have absolutely anything to show. And that is very worrying. Delays. Despite having invested billions of dollars in that restructuring of the company to bet (practically) everything on AI, three internal sources confirm that Meta finds it very difficult to meet the planned deadlines. The race for generative AI waits for no one, and at the company headquarters nerves are on edge because the roadmap is not being met. Avocado, where are you? The new foundational AI model that Meta has been working on for months has been internally named Avocado, but at the moment it is not measuring up, something that reminds us what happened to Llama 4. Internal tests reveal that although it manages to surpass the aforementioned Llama 4 and the old Gemini 2.5, it falls short of Gemini 3.0 (and of course, the recent Gemini 3.1). Patience. Coming out with a model that is clearly worse than its rivals does not make sense, so Meta has decided to wait and delay the launch of its model. Avocado is expected to hit the market in May at the earliest. And meanwhile, Gemini. The situation is so critical that according to these sources, the leaders of the AI ​​division are considering something unthinkable: paying a license to Google to be able to use Gemini in their own products, something that for example will Apple do Siri. That would be a clear sign that for now this own model is not capable enough to power the AI ​​functions of WhatsApp, Instagram and Threads. Money does not equal speed. The company has spent billions of dollars on AI researchers, and has committed to invest 600,000 million dollars in building AI data centers. In January, Meta projected a capex of $135 billion dedicated almost entirely to these projectsalmost double the $72 billion it spent last year. Despite these investments, the company is currently missing from an area in which its competitors continue to advance. Internal tension. According to these sources, Meta is becoming a tinderbox. The “TBD Lab” (for “To Be Determined”), the unit led by Wang, is working under maximum pressure on models named after fruits (Avocado, Mango, Watermelon), but has clashed with old-school Meta managers like Chris Cox and Andrew Bossworth. The company is trying to integrate those models with Meta’s advertising business, which is what supports everything, but Wang doesn’t seem to handle that part of the business very well. Goodbye to open models. Meta stood out at the beginning of this AI race as the company whose open models —not Open Source— were above the rest. Llama became the norm in this area, but in this new stage that philosophy seems to change and China is the one that now leads that segment. Thus, there is talk that both Zuckerberg and Wang lean toward closed models, such as those of OpenAI (GPT) or Google (Gemini). This allows you to have full control over the code, a competitive advantage that Meta does not seem to want to give up. Few fruits of this tree. Despite the extraordinary deployment of resources, the current balance is poor. Meta’s only tangible product of those investments is Vibes, an application similar to Sora that has not managed to fully gel. Meanwhile, those initial talent signings have turned into abandonments: the trickle of AI researchers who leave the company to join others (or found their own projects) is increasing. In Xataka | Meta has been buying chips from NVIDIA and AMD for years. Now it also makes its own so as not to fall short

8 GB of RAM has gone from $40 to $130 in five months. It is explained in four words: “It is what it is”

by

At this point in the film, it no longer escapes us that we are in the midst of a new component crisis. What started with RAM crisis ended up mutating into SSD crisis and any device that has memory or a memory controller. We are in an “unprecedented” situation, said by the companies themselves that manufacture that memory, and although it all started with SSDs and more expensive RAM ‘chips’, now things have escalated. How much? Let’s go with some examples. a rocket. Maybe in Europe they are not the best known, but Framework It is a company that is doing things well. It has a desktop PC, but also something much more interesting: modular laptops. It is not so common to be able to choose all the components of a laptop, and the Framework components do give us that opportunity. The fact is that they are the perfect example to see how the market is. In September they stopped selling standalone RAM modules. They were not the only ones who began to sell computers without memory or with less physical SSD than advertised, but now the next ‘stick’ has arrived: if before 8 GB of RAM cost 40 dollars, now it is 130 dollars. And if you wanted 96 GB of RAM, before it “only” cost you $480… and now you have to shell out $1,340. It’s the market, friend. It is estimated that prices are increasing between 6% and 16% on the company’s equipment. The Framework Desktop 32GB LPDDR5X memory is up $110 since it launched. And the 128 GB one has increased by about $600. The equipment that was already built in the warehouse has not been affected by this, but as the stock runs out, they will inevitably follow the same path. “The price is what it is, unfortunately,” says the CEO of Framework In ArsTechnica we can read that Nirav Patel, CEO of the company, points out that they are trying to solve the problems, but in the end… he is not on his own and the best thing he can do is be transparent. In an interview with BIpoints out that they are looking under the rocks and that if an intermediary tells them “we have found 5,000 RAM modules in a warehouse”, they would buy them without thinking. The problem is accessing new RAM modules… because there aren’t any. Increases. As we say, it is no longer just the RAM that we can see in stores like Amazon, but the components of a computer, a cell phone, a television, a cara modem or… one Raspberry Pi. Since this crisis began, we have talked a lot about how manufacturers were saying that things were tough and it was going to take a while for the market to recover. But the case of Framework helps ground things, and so does Raspberry. Because if Framework uses DDR5 memory, which is the most advanced, Raspberry’s is not the latest generation. However, the company has also had to raise prices. Yes three months ago increased a little, now they have skyrocketed. Its memories are LPDDR4 and the company has published a table that point Because the more RAM your board has, the more the price goes up. Raspberry PI 4 and Raspberry Pi 5 Price increase 1 GB Nothing 2GB 10 dollars 4GB 15 dollars 8GB 30 dollars 16 GB 30 dollars Out of the race. In the end, it all depends on the price of the device. In a 4,000 euro computer that you need for professional work, you have no choice but to pay about 600 more to expand the RAM. But on cheaper equipment, the feeling is that it has a much greater impact. This is the case of the Raspberry. In a 120 euro device like the Raspberry Pi 5 of 16 GB, an increase of 60 euros is stupid. And as the situation lasts a long time, that will be the big problem for many software manufacturers. Apple just present new laptops and the iPhone 17e and, in Spain at least, the price has remained the same. This has not happened in other countries such as the United States, seeing strong increases in some models. Apple, Lenovo, Dell… are companies that have already said that things are bad and the users will have to bear the cost. But there are also voices that point out that not all companies can allow their users to be the ones who ‘swallow’ with the situation. From SMICChina’s large foundry, has already pointed out that there will be hardware companies that will be left out of the game. We are already seeing it: the cheap mobile is suffering the consequences and the Chinese Meizu, which wanted to eagerly return to the Western market, is already backing down. 2027 2028. Valve and its Steam Machine is another example: the console should come out this spring, but not only is there no price, but it is not known when it will arrive. And when will the end of this catastrophic situation be? It’s the million dollar question. Patel comments that the relief will begin in early 2028, a date similar to what other parts of the industry are managing. However, Jensen Huang, CEO of NVIDIA, has already warned that the AI ​​race He has seven or eight years leftand just now they have begun to commission TSMC to begin mass manufacturing of Vera Rubin, their next-generation acceleration platform. It is something that needs memory and only Samsung and SK Hynix (two of the big three RAM companies) are able to supply it right now. In the end, it is about going day by day in this new crisis, but everything indicates that if we need something, it is better to buy it as soon as possible because ifamsung, Micron and SK Hynix they are not doing consumption memoryprices will rise more and more over the next few months. Images | Raspberry, Framework In Xataka | SK is one … Read more

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.