companies

They have kidnapped agents from Anthropic, Google and Microsoft for the sake of science. The three companies ended up paying

by

In some development teams it is already becoming common to rely on artificial intelligence agents to review incidents, analyze code changes and move through tasks that were previously left in human hands. The problem appears when these systems not only read information that may come from outside, but also operate in spaces where they coexist. sensitive keys, tokens and permissions. That is what recent research puts on the table: we are not simply facing a useful tool that can make mistakes, but rather an architecture that can also become dangerous if it is deployed without very clear limits. The alarm has been turned on Aonan Guan and Johns Hopkins researchers Zhengyu Liu and Gavin Zhong after demonstrating attacks against three agents deployed on the aforementioned platform: Claude Code Security Review, from Anthropic, Gemini CLI Action, from Google, and GitHub Copilot Agent, a GitHub tool under Microsoft. According to your documentation, The failures were communicated in a coordinated manner and ended in financial rewards paid by the companies, but what is relevant is that they point to a broader problem. This is how they managed to twist the agents from within The name that Guan gives to the discovery helps a lot to understand what this is all about: “Comment and Control.” The idea is simple to explain, although the substance is not so simple. Instead of setting up an external infrastructure to direct the attack, GitHub itself acts as an entry and exit channel: the attacker leave the instruction in a titlean incident or a comment, the agent processes it as if it were part of normal work and the result ends up reappearing within that same environment. Everything stays at home, and that is precisely the key to the problem. And that “everything stays at home” is not a minor detail, but the basis of what the research describes. The three agents share a very similar logic: they read normal content from GitHub, incorporate it as a work context, and from there, execute actions within automated flows. The clash appears because that same space not only contains text sent by third parties, but also tools, permissions and secrets that the agent needs to operate. The first case Guan details concerns Claude Code Security Review, an Anthropic GitHub action designed to review code changes and look for possible security flaws. Up to this point, everything is within what was expected. The problem, as the researcher explains, is that it was enough to introduce malicious instructions in the title of a pull requestwhich is the request that someone sends to propose changes to a project, so that the agent will execute commands and return the result as if it were part of your review. The team then managed to go a step further and demonstrate that it could also extract credentials from the environment. The interesting thing is that the same scheme also appeared in the other two services, although with nuances. At Google, Gemini CLI Action could be pushed to reveal the GEMINI_API_KEY from instructions snuck into an issue and its comments; In GitHub Copilot Agent, the variant was even more worrying, because the attack was hidden in an HTML comment that a person did not see on the screen, but the agent did process when another person assigned it to the case. In both scenarios, the background was the same again: apparently normal content that ended up twisting the behavior of the system until exposing credentials or sensitive information within GitHub itself. Guan assures that the pattern made it possible to leak API keys, GitHub tokens and other secrets exposed in the environment where the agent ran, that is, just the credentials that can later open the door to much more delicate actions. Who does this affect? Especially to repositories that run agents in GitHub Actions on content sent by untrustworthy collaborators and, in addition, give them access to secrets or powerful tools. The researcher himself clarifies that the risk depends a lot on the configuration: by default GitHub does not expose secrets to pull requests from forksbut there are deployments that open that door. And here another layer of the matter appears, less technical but just as important. As published by The RegisterAnthropic, Google, and GitHub ended up paying bounties for the findings, but none of the three had published public notices or assigned CVE at the time of that information. Guan was quite clear about this: he said he knew “for certain” that some users were still stuck on vulnerable versions and warned that, without visible communication, many may never know that they were exposed or even being attacked. So although there were mitigations and changes in documentation or in the internal treatment of reports, there was no equivalent public notice for all those potentially affected. Anthropic settled the case on November 25, 2025 and paid $100 Google rewarded the discovery on January 20, 2026 with $1,337 GitHub closed the case on March 9, 2026 with a payment of $500 What makes this case especially delicate is that GitHub does not seem like the end of the road, but rather the first visible showcase. Guan argues that the same pattern can probably be reproduced in other agents who work with tools and secrets within automatic flows, and there he mentions from Slack-connected bots to Jira agentsmail or deployment automation. The logic is the same again: if the system has to read external content to do its job and also has enough access to act, the field is fertile for someone to try to twist it from within. The conclusion that Guan reaches is not about selling a magic solution, but about returning to a fairly classic idea in security: giving each system only what is essential to do its job. If an agent reviews code, they shouldn’t have access to tools or secrets they don’t need; If you’re just summarizing issues, it wouldn’t make sense for you to write to GitHub or touch sensitive credentials. That … Read more

Europe thinks that it is the one who wants to become independent from US technology companies. It’s actually the other way around.

by

We are going to have to make a dictionary when we talk about artificial intelligence. Yes generative artificial intelligence, general artificial intelligence, AI agents and Google has its Personal Intelligence. This service has been available for a few months in the United States and is now expanding to the rest of the world for users of the company’s suite. To all? Well no, not everyone, and Google leaves out the entire European economic area, Switzerland and the United Kingdom. What does this do. First of all, Gemini Personal Intelligence is like an ‘entity’ that has an eye on all the Google applications that we use with our account. It is a connection between the entire ecosystem which has access to all the information from YouTube, Maps, Calendar, Drive, Gmail, Docs or Photos. The idea is that they know everything about you to help you with day-to-day requests. Google itself gives an example: Since I connected my apps through Personal Intelligence, my daily life has become easier. For example, two weeks ago we needed new tires for our 2019 Honda minivan. While waiting at the store, I realized I didn’t know the size of the tires, so I asked Gemini. Nowadays, any chatbot can find these tire specifications, but Gemini went further. He suggested two options: one for daily driving and one for all-weather conditions, with references to our family road trips to Oklahoma that I found on Google Photos. Then, he neatly extracted the ratings and prices of each one. When I got to the counter, they asked me for my license plate. Instead of looking for it or losing my spot in line to get back to the parking lot, I asked Gemini. It extracted the seven-digit number from an image in Photos and also helped me identify the specific model of the truck by searching Gmail. Just like that, we had everything ready. Everyone can appreciate how useful this is, of course. Because, furthermore, it is not free: you need to be on an AI Pro or Ultra payment plan. Europe, you are excluded, beautiful. But if you think that life can be solved for you like the person in the example and you live in Europe, you should know that you cannot access it. Google has made it very clear that the feature is not available in the European Union, the United Kingdom and Switzerland due to the strict privacy regulations that are found within the General Data Protection Regulation. It is one more service that does not reach European users due to these stricter privacy conditions and, in addition, they have not detailed any deadline so that we Europeans know more or less when it will be available in the region. If you read us from a Latin American country and you are interested in this software, good news, is available. Privacy. Regarding privacy, being something that we should take very seriously, it is curious how in the age of AI that privacy is eroded because the models have more and more data. We give a lot to this software and we don’t really know who’s watchingand Google wanted comment What is the privacy approach of your Personal Intelligence. According to the company, we can link and unlink applications from the ecosystem according to our preferences, but once we connect Photos, for example, you will have access to everything. They say that the photos in the gallery will not be used to train the model (here it is extremely sensitive because we each have personal photos on our mobile) and they take “measures to filter or hide personal data from the conversation.” They point out that “we do not train our systems to learn your license plate number, but to understand that, when you ask for it, we can find it.” Regarding health, a sensitive topic that is topical due to what El Salvador has done, they assure that “Gemini tries to avoid making proactive assumptions about sensitive data such as your health, although it will analyze that data if you ask it.” ¿Pressure? In any case, and what Google’s Personal Intelligence has to offer draws more or less attention, it is evident that carrying out a global launch of this magnitude without waiting to ‘sort the papers’ to launch it in Europe is also a declaration of intent. Europe is at a time when it is seeking its military sovereignty, aerospace and technologicalareas in which depends largely on the United States. We are building our infrastructure and systems, and not launching something like this in Europe is one more way of putting pressure on the organizations that, very actively, sand have positioned about the treatment of our data by these large companies. Image | Google (edited) In Xataka | Sometimes erotic AIs are AIs. And sometimes, they are a man from Kenya who charges two dollars an hour

Chinese companies are designing their AI for an audience the West is ignoring: retirees

by

We can adapt the title of that great Cohen film to AI: there is no AI for old people. The majority of AI chatbot users are young and the older ones usually have technical knowledge. The AI ​​boom, like other technological booms, is leaving out the older onesexcept in China. Hello, grandmother. They tell it in Nikkei Asia. Large Chinese technology companies such as ByteDance and Tencent are designing chatbots and apps with AI with older users in mind. Doubao, the most popular chatbot in China, has launched advertising campaigns targeting retirees, highlighting its accessible features; It allows you to converse by voice, understands dialects and even addresses users as grandfather or grandmother. According to data from China Internet Network Information Centerthe number of AI users between 50 and 59 years old represent 10% of the total and those over 60 years old only 5%. They are still a minority, but there is a curious fact and that is that, although the adoption rate in this group is much lower, the users who start using it are more loyal and use it frequently. Everyday help. In Nikkei they tell the story of Chen Bing, a 63-year-old woman who has made AI her personal assistant. He used it to organize an event with alumni of his school, from sharing expenses to generating a video that he used in the background of a poetry workshop. It also helps you identify flowers and read fine print. According to Chen, AI gives him independence and prevents him from having to constantly ask his children for help. And health. There are other AI proposals aimed at the elderly, such as Ant Afu, a health chatbot with which users can get advice and access health services. However, it has generated criticism, first of all due to possible conflicts of interest. In the past, there was a scandal because Baidu recommended hospitals and treatments based on paid advertisements and there are doubts that this system has similar influences. On the other hand, there is the question that AI continues to fail a lot in diagnosis. The silver economy. It is what the market for products and services aimed at older people is called in China. China already has 323 million retirees and the government is promoting these types of initiatives since it sees great potential for consumption by the elderly, something they need to encourage in the midst of an economic recession. It is estimated that by 2035, the silver economy will account for 10% of the country’s entire gross domestic product Aging population. It is one of the problems facing China today. The government is trying literally everything for stimulate birth (without much successby the way) and have also raised raise the retirement age. However, the aging of the population is not something exclusive to China, it is also a problem that Europe and more countries in the northern hemisphere We have been dragging on for a long time. In the European Union there are some initiatives such as digital literacy courses for seniorsbut at the private company level, the proposals are very niche. In Xataka | China knows that its population is going to collapse but it already has a long-term plan to solve it. Of course, thanks to AI

LaLiga’s massive IP blocks are making life impossible for users, companies and developers. So you can claim

by

LittleCranky67, which is the alias of our protagonist, didn’t know what was happening with his computer this weekend. This developer was doing something that never gave him any trouble: working with the GitLab platform to download a Docker software package. That process kept giving him strange errors, and LittleCranky67 ended up realizing what had caused it all: LaLiga’s indiscriminate IP blocking. After share your frustration on HackerNewshundreds of comments confirmed other similar cases, and in them we also discovered something interesting: how to officially claim LaLiga. Or at least, how to try. A sad old story. LaLiga he shields himself in the Judgment of December 18, 2024 issued by the Commercial Court No. 6 of Barcelona. This allows you to demand from operators such as Movistar, Vodafone, Orange or Digi to block at the IP level any address that is identified as a source of illegal IPTV broadcasts during LaLiga football matches. Many of those IPs are Cloudflare shared IPs, so when the IPTV service IP is blocked, all domains associated with that shared IP are blocked, which can be hundreds or even thousands. And in those domains there are web pages of private usersfrom companies that they stop being able to sell and also critical services for developers such as Docker, GitHub or GitLab. The irony is that lockdowns don’t work. While many users complain about these blocks and how are affecting websites and services that usemany others continue to remember on social networks that in reality the blocks to view these IPTV broadcasts can be easily circumvented in many ways. The most popular, use VPN services. In LaLiga they know that this method is widely used, so for months They are also working on blocking those services. It doesn’t seem to be serving a lotand whoever really wants to watch the football game without paying has many relatively simple ways to achieve it. If you are affected, you can claim. In that thread, several users remember that one way to try to change things is for users to protest, complain and complain en masse. There are several ways to do it: Telecommunications User Service Office. It is the official body in Spain for these cases. A formal claim can be filed for arbitrary loss of service or censorship, and even claim financial losses if the blockade prevents you from working. Those who have a digital certificate or Cl@ve can do so directly online. Complain to your internet provider. It is also important to open a support ticket with your operator. It is true that they are obliged to follow court orders, but they must know that the blockade is causing collateral damage to services that have nothing to do with football. Common Electronic Registry (SARA network). This portal It also allows you to send formal complaints to management if other methods fail. Spanish Data Protection Agency (AEPD). Those responsible for RootedCON have been fighting this situation for some time, and offer another recommendation: report LaLiga to the AEPD. This template allows you to complete that complaint in a simple way Demagive to Telecommunications Operators. At RootedCON they also suggest filing a complaint against ISPs, and explain the process in a small thread on Twitter. Again, just download a request and file it individually. Complaint to the European Commission. It is also possible to enter the European Commission complaints website to send a claim to the entity. We explained it in Xataka and the process is another way of trying to stop this situation with the help of European bodies. The BOE serves as a defensive argument. In these complaints it is advisable to cite the BOE-A-2022-10757 as a legal reference. It corresponds to Law 11/2022, of June 28, General Telecommunications (LGTel) and is the fundamental rule that regulates your rights as an internet user in Spain. The message that we can write is the following: “Under the protection of Law 11/2022, of June 28, General Telecommunications (BOE-A-2022-10757), specifically regarding the rights of end users (Chapter IV) and the principles of continuity and quality of service, I present this claim for the blocking of access to legitimate IP addresses (specify which ones, e.g. Cloudflare/Docker) unrelated to any illicit activity. This blockade constitutes a violation of my right to communication and the contracted service, causing harm (professional/personal) by preventing the operation of work/security tools. “I request the immediate cessation of said technical restriction in compliance with the provisions of the aforementioned Law.” The nightmare continues. The debate in HackerNews is nothing more than confirmation of what internet users in Spain have been suffering for more than a year. A private organization has the power to order ISPs in a country to indiscriminately block IPs without judicial review in real time, during regular hours, causing documented harm to third parties that have nothing to do with the original violation. In that thread some users compare the situation with that of the Great Firewall of Chinanot so much in intensity as in its logic. We are faced with an infrastructure of selective censorship that seems to be able to be applied to any content that an actor with sufficient judicial power wants to block. From football to tennis or golf. In fact, things could go further, because what began as an attack against illegal broadcasts of football matches could now be seen in other sports such as tennis or golf. Telefónica —which follows in the footsteps of LaLiga— wants to extend indiscriminate blockades to the Champions League, tennis or golf. This threatens to suffer these side effects for many more days and for many more hours, and can mean that for a good part of the week, users like LittleCranky67 find themselves unable to download Docker packages or access thousands of legitimate websites that end up being knocked down by these blocks. Images | Wirestock | LaLiga In Xataka | LaLiga has been at war with Cloudflare for years over piracy. It has just joined forces with its main competitor

with seven Dutch companies together

by

These American companies control 60% of the global cloud infrastructure market: Microsoft and its Azure, Google Cloud and Amazon AWS, a providential sector in the globalized and permanently connected world in which we live. And very lucrative: in 2025 revenues exceeded $400 billion, according to Synergy Researchthis is nine times more than in 2017. There are no corporations capable of overshadowing these three, so seven Dutch cloud services companies they have made a decision: unite to be a real alternative to American big tech. The movement is more important than it seems: it is an organized response to such dependency that it is already considered a strategic risk. The project: Open Cloud Alliance. The answer is Open Cloud Alliantie, a conglomerate formed by Centric, KPN, Info Support, Intermax, Nebul, Previder and Uniserver, with a joint turnover of 2.5 billion euros annually. In their manifesto they explain that they are creating jobs in the Netherlands and that both companies and those who work in them pay taxes there. As explains Ludo BaauwCEO of the Intermax Group to NRC, separately they are competitive and their reason for being is not to set prices, but to bid for public contracts: “I would prefer a competitor from the Netherlands to win rather than a large American technology company.” There was a trigger to join: the possible sale of Solvinity, provider of cloud services for the Dutch government’s Digid digital identity system, to the American company Kyndryl. The agreement is pending approval by the Ministry of Economic Affairs, but it has already had consequences. A strategic vulnerability. The first consequence was to put on the table a vulnerability of the Dutch system that can be extrapolated to all the states of the old continent. According to an analysis carried out by NOS, 67% of the domains of Dutch public organizations, hospitals and schools depend on at least one American cloud service. Why is it important. The project has three compelling reasons for existing: Be real competition from North American big tech. The CEO of the Dutch competition body (ACM) has made it clear: “Overall, alliances like this can boost market forces by creating new players that are better positioned to compete with large American suppliers.” Boost to the national economy. The companies are clear in their manifesto: jobs and taxes for the Netherlands. In a phrase: “it is not an expense, it is an investment.” Data sovereignty. That such critical state services as health, education or digital identity depend on foreign companies subject to foreign legislation and corporate decisions outside of European control. Context. This movement arises within the European debate on digital sovereignty and reducing technological dependence on the United States. The trend is not new, but Trump’s policies have accelerated that conversation. Europe has the legal framework in the form of GDPRthe Digital Markets Lawthe Digital Services Law wave Chips Actwhich make up a solid regulatory arsenal aimed at reducing foreign technological dependence. The problem is that having the laws is not equivalent to having the industry. Local European suppliers are individually solvent, but they do not have the capacity to absorb complex projects or compete with the scale of the big three that dominate the market. Not even GAIA-Xthe large Franco-German sovereign cloud project, has been able to so far. Europe regulates well but scales poorly and that is the void that the Open Cloud Alliantie is going to try to fill. How are they going to work?. He operating model It will be based on three pillars: Common technical standards, which will allow data to be moved between providers without friction by adopting the same technical specifications. Collaboration yes, cartel no. They will share standard infrastructure and may bid together for large contracts, but they are still competing with each other when it comes to winning customers. Sovereignty clause. If one of the seven is acquired by a non-European company, the others automatically absorb its role. The data always remains in Dutch hands, regardless of what happens in the mergers and acquisitions market. Towards technological sovereignty of the cloud. The Open Cloud Alliantie is a relevant experiment on which other member states and corporations will set their sights as long as it is perfectly replicable. Medium-sized companies that otherwise could not compete with large companies in the United States, but that, grouped under common standards and clear collaboration rules, can offer a credible alternative to the public sector. The question is whether other European countries will take note before the dependency becomes too deep to reverse. In Xataka | Europe is looking for a place to put its AI gigafactory. Spain and Portugal are showing all their renewable plumage In Xataka | Europe has proposed to become technologically independent from the US: And it has started with the most difficult thing: chips Cover | İsmail Enes Ayhan and François Genon

There are companies that want to install an age verifier

by

Already in 2018 we were talking about vaping was becoming popular among teenagers and today, eight years later, things do not seem to have improved. In schools it has become a recurring problem with students carrying their vaper in their backpack as if it were just another accessory. Now there are companies looking for solutions so that minors cannot access their products. Age verification. The concept is very topical in the technological world. In the midst of the debate about whether prohibit access to social networks for those under 16 years of ageage verification systems are presented as a critical element to make this possible. What we did not expect is to find this concept linked to vapers or electronic cigarettes, but it is exactly what some manufacturers of this type of products are considering, as they say in Wired. Biometrics and blockchain. It is the proposal of IKE Techa company formed from the collaboration between the vaper manufacturer Ispire Technology and Chemular, a consulting firm specialized in the nicotine market. Their approach is to use a combination of biometrics, blockchain and a BLE chip integrated directly into the cartridge to guarantee the user’s age. The system works through an app in which the identity document is scanned and a selfie video is taken. This data is verified using an identification service such as Clear or ID.me and, if the verification is positive, the vaper is unlocked via Bluetooth. For it to remain active, the mobile phone with which the identification was made must remain close to the vaper. According to the company’s tests, the system showed 100% effectiveness. The reason. It is not that vape manufacturers have suddenly decided to insure their products, but it is a consequence of regulation. At the beginning of the month the FDA (the body that regulates foods and substances in the US) published a warning draft of the risk that flavored electronic cigarettes pose for young people. Although at the moment they are guides and their application is not mandatory, it makes it clear that this is where the regulation is moving, which is why there are already companies taking steps to protect themselves. Doubts. IKE Tech claims that it showed its technology to the FDA and they loved it, but there are many doubts about it. Speaking to Wired, Stanton Glantz, director of the Tobacco Control Research and Education Center, is not clear that this system will really work and believes that “every technical solution has a way of getting around it.” The truth is that once the vaper is activated, the system does not control if another person is using it, so an adult could activate it and then a minor use it. For Wang, a system that would work is to install geofences so that vapers are deactivated near schools or in prohibited places such as airplanes. Legal loophole. The fact that vapers have been so accessible to younger people is that there was a legal vacuum that did not equate them to tobacco in terms of regulation. He draft anti-smoking law of 2025 finally put them at the same level, which means that smoke-free spaces also include them and their consumption is prohibited for those under 18 years of age. The law is expected to come into force later this year. Flavors and colors. There is also another issue that has attracted the younger audience and that is marketing. Organizations like the WHO wave AECC have warned that vapers have been marketed in countless fruit or candy flavors, with colorful designs and a fresh and striking aesthetic, almost as if they were toys. In addition, strategies such as advertising at festivals or through influencers who held contests and raffles have been used. To this we must add the fact that there is no specific legislation, which means that we find disposable vapers in all types of stores within reach of anyone. According to data from the 2025 Secondary Education Drug Use Survey50% of young people between 14 and 18 years old claimed to have tried vaping and 27% admitted to having done so recently. The problem is still there. Even if the law comes into force and the age verification systems work perfectly, the question remains that They are very harmful to healthwhether they have nicotine or not. When vaping, the lungs are exposed to a wide variety of chemicals such as propylene glycol, vegetable glycerin or flavorings. The argument for vaping is that it is better than the combustion that occurs when smoking traditional cigarettes and there are studies that suggest that this is sothe question is whether replacing an evil with a lesser evil is the solution. In Xataka | The tobacco industry had found an escape route in vapes. And Spain is already considering putting an end to it Image | VapeClubMY in Unsplash

fewer and fewer companies offer it

by

When the pandemic forced millions of Spaniards to work from home, many thought that nothing would be the same again. Companies had shown that it was possible and employees had found that they would rather do it than lose hours of your life in traffic jams or by public transport to their offices. And yet, six years laternumbers tell a story completely different. The 16th edition of the report State of the labor market in Spain 2025 that InfoJobs and Esade have just presented makes it clear: teleworking has not only not grown in new job offers, but it has been declining for four years in a row. And in 2025, remote work has hit a low that few anticipated. A setback that is no longer punctual. The study indicates that, in 2021, when hiring for remote work was at its peak, 21% of the vacancies published on InfoJobs included some form of remote work. In 2025, that figure dropped to 11%, which represents 280,810 positions with this modality. Four consecutive years of decline already make this data a structural trend, not a temporary correction in business culture. The sharpest drop in this type of day It occurred between 2023 and 2024, with a decrease of four percentage points in a single year. In 2025, three more points were lost, leaving the data below even the teleworking offers that were published in 2020, in the midst of the health crisis. Mónica Pérez, director of communication and studies at InfoJobs, assures that “many companies that launched into these models and saw the B sides related to productivity, team building or cohesion, backed off somewhat.” What the official data say. It should be noted that the data provided by the Infojobs study refers to new job offers, not to positions that already existed and have been maintained during these years. When observing both phenomena we find opposite trends. As companies cut teleworking options in their new vacancies, the data from the Active Population Survey (EPA) highlight that the real number of employees who telework in Spain has continued to grow. The last ones consolidated data of 2024 indicate that 14.6% of employed people teleworked regularly or occasionally, the highest figure high since the pandemic. Of course, Spain is still very far from the European average. According to EurostatIn the European Union, 22.6% of employees carry out some form of teleworking (100% remote or hybrid workday), compared to the 15.4% assigned to Spain. Countries such as the Netherlands or Sweden exceed 46% teleworking, which gives an idea of ​​the existing gap and to what extent the in-person work model continues to be dominant in the Spanish labor market. Sectors that make a difference. The gap between sectors is also enormous, although the nature of each job makes it logical. IT and telecommunications is the most favorable field for remote work, with 68% of its vacancies including the teleworking option. This is followed by the legal sector with 59% and the finance and banking sector with 51% of vacancies with teleworking. At the opposite extreme are the sectors where presence is simply inevitable: logistics and warehouse, health, tourism and restaurants or retail sales. The study hardly sees any differences in the number of vacancies with teleworking with respect to the size of the companies, which indicates that it is a factor that depends more on the sector to which it belongs than on the number of workers that form it. Madrid and Barcelona, ​​​​in another league. The geographical distribution teleworking is not homogeneous either. Madrid reaches 40% of all vacancies with some remote modality published in 2025. Catalonia follows, with 19% of vacancies with teleworking, followed by Andalusia with 11% and the Valencian Community with 7%. There is a quite logical reason behind these data: Madrid and Barcelona concentrate most of the corporate headquarters, consulting firms, offices and technology companies in the country, the sectors with the greatest compatibility with remote work. This pattern coincides with what the INE already recorded after the pandemic, when teleworking skyrocketed especially in Madrid and Catalonia compared to the rest of the communities. In Xataka | Working from anywhere was the dream of teleworking: not notifying those location changes can get you fired Image | Unsplash (Rodeo Project Management Software)

OpenAI has signed countless billion-dollar agreements with other companies. We are discovering that they are made of paper

by

OpenAI has announced that will abandon development of Soraits AI video generator, just six months after the launch of its standalone app. Disney, which had announced a $1 billion investment in OpenAI in exchange for licensing its characters for Sora, has confirmed that the deal will not go ahead. The money never changed handsand joins others in recent weeks that send a worrying message. One that calls into question the real strength of the most valued company in the AI ​​sector. Paper agreements. In recent months, OpenAi has been the protagonist of a frenetic string of announcements that have shaken the stock markets and sent prices skyrocketing. Analysts like Ed Zitron have documented in detail how these agreements are for now more smoke than anything else: all of them were “letters of intent”, conditional commitments that now seem increasingly difficult to come true. There are examples everywhere. The NVIDIA case: the one hundred billion that did not exist. In September 2025 NVIDIA announced a “strategic partnership” with OpenAI to invest “up to 100 billion dollars” and build 10 GW of data centers. Four months later, the company led by Jensen Huang considerably reduced that investment to 30 billion dollars. Jensen Huang recently stated that this will “probably” be the last round he will enter into OpenAI and clarified that the statement made it clear that this was a “letter of intent”, not a contract. Months later in NVIDIA’s quarterly results, the agreement is described as “an opportunity to invest in OpenAI.” Not a single dollar has been sent to him, and it is not certain that he will. The AMD case: 34% rise in the stock market. In October, another mega-deal. amd announced a “definitive” agreement with openAI to deploy 6 GW of data centers. The company indicated that would potentially generate “tens of billions in revenue,” and AMD shares rose 34% in one day. Four months later, in quarterly results from the company, zero mentions of OpenIA. IN November 2025, in AMD’s 10-Q filing, AMD’s outstanding obligations on contracts with a duration greater than one year were 279 million dollars. There were practically no mentions of OpenAI. Many promises, no reality. The Broadcom case: a confusing order. Broadcom too was going to deploy 10 GW of “AI accelerators designed by OpenAI” at the end of 2029, but at the moment there is still no evidence that chip sales have occurred and there are no clues in OpenAI’s latest quarterly results, which do not mention this agreement anywhere or its impact. Broadcom CEO he did tell investors that they expected to deploy 1 GW of computing in the form of XPUs in 2027, but did not give details of how they planned to reach 10 GW in 2029. And also revealed that “we do not expect much in 2026” from the contract with OpenAI, because the return will focus on 2027, 2028 and 2029. The Disney case: a very bad sign. The agreement with Disney announced in Decemberincluded the company taking a $1 billion stake and will license more than 200 characters from Disney, Marvel, Pixar and Star Wars for use on Sora. It was the type of agreement that validates a company before the general public, especially since Disney does not sign agreements with just anyone. However, the agreement was entirely built on stock warrants, not cash, they point out in Deadline. By abandoning Sora, Disney has withdrawn without consequences and without having transferred a dollar. Another paper agreement. The SKHynix case: where are we going to get so much memory from?. SK Hynix and Samsung intended to provide 900,000 RAM wafers per month for OpenAI’s Stargate project, but the result of these intentions has been null. That agreement would have consumed 40% of world production of DRAM in the midst of the crisis of this type of components. The mysterious Norwegian data center case. OpenAI promised in July 2025 that would boost construction of an AI data center belonging to the Stargate project but which would be in Norway. It was then expected that this center would have 100,000 NVIDIA chips by the end of 2026, and that it would expand “significantly” from that figure. There has been no news of this development since then. Nobody asks questions. Zitron complained in your reflection how financial analysts seemed not to ask the necessary questions when faced with these announcements. He explains that OpenAI had committed about $300 billion in different agreements to create new data centers, but its real income is around $4.5 billion a year and it is expected that it will have losses of about $14 billion in 2026. Despite everything, Zitron criticizes, the stream of advertisements continues to work because it generates increases in the stock market and positive headlines. The difference between contracts and letters of intent was buried in the fine print of the advertisements that almost no one reads. And the examples continue. In fact, the advertisements do not stop coming despite everything and everyone. OpenAI announced in February an investment of 110 billion dollars by SoftBank (30 billion), NVIDIA (30 billion) and Amazon (50 billion). SoftBank itself is “testing its lending limits” with that bet, which we will see if he can complete. Amazon’s 50 billion are divided in two phases: a first of 15,000 million that should be executed on March 31, and another of 35,000 million dollars whose deadlines depend on several events. Too many agreements that must demonstrate something critical: that they are not made of paper. In Xataka | Problems are multiplying for OpenAI in the race for AI. Your solution: go from 4,500 to 8,000 workers

China is giving apartments to its entrepreneurs because it is clear that the future is the Yo SL companies

by

Ma Ruipeng is 41 years old and has been working as a programmer for 20 years. Three months ago he left his job to start his own company. From his apartment in Beijing he works with three computers, AI tools like Claude Code, design platforms like Figma and, of course, his own installation of OpenClaw which he has called “Big House”. That’s what he hopes from his solo adventure: that his house becomes really big. He hasn’t made money yet, but he clearly prefers working with AI before AI works in place. The era of the Yo SL in China they start to push There are increasingly so-called “one-person companies” (OPC), one-person companies that act like startups founded and operated by a single person. These types of entrepreneurs make the most of AI tools—scheduling agents, video and image generators, task automation systems—to do the work that previously required having a team of employees. The falling cost of developing digital products, combined with the arrival of AI agents really functional like OpenClaw has made this type of business figures viable for the first time on a massive scale. The government is betting on entrepreneurs in the AI ​​era. In November the city of Suzhou advertisement that would build “30 OPC communities” with the goal that by 2028 the city would have at least 1,000 one-person AI companies. Other Chinese cities quickly followed. The Pudong district of Shanghai covers up to 300,000 yuan (37,500 euros) in computing costs, and Wuhan offers special loans for AI solopreneurs and even promises to absorb some of the losses if they go bankrupt. It is a well-known strategy: there is a central guideline that drives core competence to take advantage of this new industry that promises to revolutionize the market. Free floors and empty data centers. Chinese government incentives they don’t just translate into money. Several local governments are converting office buildings and underutilized data centers in a kind of incubators for this new SME format, for these “Yo SL”. The context is revealing, because with the AI ​​fever many municipalities built data centers without calculating real demand and had them half empty. Filling them with subsidized startups solves two problems at once. Silicon Valley is something else. On the other side of the Pacific, it is venture capital funds that finance Silicon Valley entrepreneurs, and there We bet on startups with the most return potential. In China, it is the State that is fully involved in this effort: it offers subsidies for infrastructure, it is a priority customer of these products —what’s happening with robotics— and promotes competition between municipalities to attract talent. “It’s like a giant Silicon Valley,” explained Lin Zhang, a researcher at the University of New Hampshire: “when a new technology emerges, the entire bureaucratic system is mobilized to develop it.” There will be many who fail. The uncertainty, however, is notable. Venture capitalists say that most OPCs will not end up becoming viable businesses, although they admit that government subsidies are encouraging more and more people to start pitching ideas for their startups. Taking into account that frequent layoffs are beginning to occur in the market, this is an alternative for many former employees of technology companies, who can thus seek their own opportunity with the help of the Chinese government. It is a commitment to volume as an innovation strategy: many will fail, but the more they try, the more options for success there will be. Fear of unemployment is a powerful ally. Behind many of these stories there is a common motivation: the fear of being left out of the labor market. The prospect of being replaced by AI both in China and the rest of the world is starting to get really disturbing for a whole generation of skilled workers. These OPCs are for many of them a response to that threat: if you can’t beat the AI, use it. Before, those who ended up laid off looked, for example, at franchised businesses and they set up a bar or a photoepilation business. The future indicates that many will now set up their “Yo SL”, their startup from home in which there will be no need for an office or employees. AI will take care of (almost) everything. Image | Blackcreek Corporate In Xataka | To dominate chips, China must first obtain hyper-specialized technology in the hands of its historical rival: Japan.

The US is discussing whether to condemn technology companies for designing something as addictive as slot machines: doomscroll

by

If you have Instagram or TikTok, you’ve probably been caught up in that constant river of videos, each one funnier and more interesting than the last. You like them, you share them, sometimes you comment and you continue seeing more. Without realizing it, an hour has passed. It is the phenomenon of doomscrolling and that is the reason why Meta, TikTok and Google have sat in the dock in the US. Now, the jury’s verdict is coming. The accusation. It all started with the complaint of Kaley, a 20-year-old girl, who accused Instagram, YouTube and TikTok of having designed their products to encourage addiction which ended up harming her mental and physical health as a child. He claims that one day he spent 16 hours on Instagram. Now, a jury decides whether his addiction was his fault or the design of these social networks; infinite scroll, autoplay and algorithms expressly designed to trap us for as long as possible. Why it is important. It is not the only complaint about the effects of social networks on mental health (they say on BBC that there are more than 2,000 similar lawsuits), but Kaley’s has become a reference case for being the first to reach court and also with a jury. The trial has been compared to the one that put the tobacco companies on the bench at the end of the 90s, it now remains to be seen if it has real consequences. The defense. During the trial, internal Meta documents were provided in which some employees joked that Instagram was a drug and they were dealers. However, the platforms defend themselves by arguing that each user is responsible for their own use. The director of Instagram, Adam Mosseri said at trial that social media is not “clinically addictive,” and compared it to being addicted to a television series. In addition, they defend that they have implemented safety features, such as screen time limitations and rest reminders. And now what. The platforms have been spared other accusations thanks to Article 230 of the Communications Decency Law, which exempts them from responsibility for what users publish on them. However, the lawsuit tries to get around this limitation by focusing on the design and not the content. If it succeeds, it will set a precedent and open a path for the thousands of lawsuits awaiting processing. Still, it may not be enough for real consequences to occur. In statements to, New York Times Glenn Cohen, a professor at Harvard Law School specializing in new technologies, says that even if the jury agrees with him, “it will not survive an appeal.” Chip change. In recent years, the discourse of rejection of social networks has been growing (although its use has not decreasedparadoxically) and its effects on our mental health, especially that of the youngest. Australia has banned the use of social networks for those under 16 years of age and there are other countries that have shown themselves inclined to follow in their footsteps, such as Denmark either, recently, Spain. In Xataka | Spending all day scrolling on Instagram or TikTok has a very specific effect on your brain: it dwarfs Image | Wikipedia

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.