browsers

AI browsers arrived with the promise of changing everything. Not only have they not done it: they are a danger

by

Artificial intelligence browsers did not come promising a smarter tab or a search engine with better answers. They arrived with a much greater ambition: OpenAI talks about getting closer to a “true super assistant”, Perplexity summarizes Comet as “the browser that works for you” and Google presents Gemini in chrome under the idea of ​​a new era of navigation. The promise is clear: that we stop moving around the web alone and start delegating part of the way. The problem is that that same promise is beginning to show a more delicate side. The warning. The University of Washington has now focused on this emerging risk. In an investigation Presented at the Agents in the Wild workshop, and released by the university itself on June 30, a team analyzed seven popular agentic browsers to see how they relate to basic protection of the modern web: the same origin policy. Their conclusion was clear: four of them opened relevant risk avenues and the researchers managed to run a complete proof of concept in ChatGPT Atlas in Agent Mode. The bottom jump. A traditional browser shows us pages and waits for our decisions. We can open a service, copy data, paste it somewhere else, compare options or fill out a form, but each step depends on us. Agentic browsers alter this logic because they incorporate systems capable of interpreting what appears on the screen and moving forward within the browser itself. We are no longer just talking about summarizing a page, but rather coordinating tasks between tabs, operating on open pages and completing actions that were previously left in the hands of the user. A new front. The risk does not arise just because a page is malicious, but because the agent can interpret that page as part of its instructions. That’s where he comes in prompt injectiona technique in which external content tries to alter the behavior of the model with hidden, camouflaged or simply inserted commands where the user does not expect to find them. In a chatbot, that is already a problem. In an agentic browser, the scope changes, because the system can process information from a page and convert it into actions within the browser. The barrier that was there. The same origin policy is one of those protections that we rarely see, but that underpins much of the modern web. Its function, simply put, is to prevent one page from freely reading or manipulating information from another just because both are open in the browser. Thanks to this separation, any website should not be able to simply access what we have in a bank, an email or a private service. The problem arises when an agent groups together information that was previously much more separated. Let’s imagine that we visit a seemingly normal page and ask the agent to summarize it or help us complete a task within it. Under certain conditions, that page may include content from another source, such as an iframe, along with a malicious instruction intended for the model and not for us. If the agent has sufficient permissions, it could access content that the attacking website should not be able to read directly and move some of that information to a form controlled by the attacker. The web would not have directly broken the barrier; he would have used the agent as a bridge. The important nuance. It should be noted that the study does not say that all users will suffer an attack or that any browser with AI is insecure by definition. The researchers analyzed specific versions, at a specific time, and worked with proofs of concept, not with attacks against real services or with sensitive user data. They also observed relevant differences between products: browsers that granted fewer permissions to the agent tended to reduce risk. The paradox. These browsers are attractive because they promise to save steps, understand pages, relate information and execute tasks with less intervention from us. But that same capacity is what makes the failure weigh more: it does not occur only in an isolated tab, but in an environment where there may be open sessions, personal data and pending actions. They may not yet be a massive habit, but the security debate is already here, precisely because their proposal consists of giving them more margin. Images | Xataka with Nano Banana In Xataka | Selecting all the traffic lights is no longer enough to prove that you are not a robot. Now you have to scan QR codes

Commodore has launched a retro smartphone that blocks browsers, Instagram or TikTok. It is garish, original and expensive

by

My first computer was a Commodore C64, so you will understand that this news hits a nerve with me. The brand, purchased last year by YouTuber Christian “Peri Fractic” Simpson, has now risen from the ashes with the surprising release of Commodore Callback 8020. We are looking at a smartphone with a retro aesthetic that also draws attention for its focus: the idea here is not to compete with the iPhone or the best Android phones, but rather help us not be glued to the cell phone so much since the telephone is above all that: a telephone. Commodore Callback 8020. This launch wants to take advantage of this growing trend of some manufacturers of resurrect “dumb phones”: cell phones that are capped as standard so that we can detoxify a little of that addiction that we usually have to screens. And to do this, this peculiar device does something mercilessly: block access to all types of applications and platforms Neither TikTok nor Instagram. The Callback 8020 has internet access, but by default it blocks both access to web browsers and social networks. It does so according to the company’s announcement “at the system level using a technology that is pending patent.” The device is therefore not a friend of Instagram or TikTok, but it does allow access to other internet services, such as maps or reading QR codes. Banned browsers. The brand also has its own app store, called Commostore, and to limit access to unwanted applications they have a white list, and as Simpson said at Ars Technica“social networks and browsers will never enter that white list.” Nothing about doomscrolling. Not only that: they cannot be installed using common installation methods either. sideloading. Although the current owner of the brand states that other applications can be installed through this process, this will be blocked in the case of social media apps or browsers. According to Simpson, “we’ve drawn a clear line in the sand around any app that is aimed at doomscrolling“. To avoid this at all costs, the firm indicates that they have blocked access to certain services at the DNS level “so even if you manage to install TikTok, you will not be able to access its servers.” Sailfish OS as heart. We are facing a device that uses Sailfish OS as its operational system. This development, created in 2012 by the company Jolla, allows you to install and run Android applications (99%, says Simpson), although its focus is privacy. Thus, it is perfectly possible to use tools such as Spotify, Signal or WhatsApp. The bet is completed with a very nice detail: Commodore includes some games from the C64 era to be able to enjoy them on this mobile. Limited but functional. The mobile’s specifications sheet is modest: it has a MediaTek Helio G81 SoC, 4 GB of RAM, 64 GB capacity (with microSD slot), 3.25″ internal IPS screen (480×640), 1.77″ external VFD screen, as well as a “Dome-LED” notification system. Connectivity is resolved with LTE, Wi-Fi and Bluetooth support, and we have a 48 MP rear camera and an unspecified front camera that does have autofocus. The battery is 1,550 mAh and recharges via USB-C. Retro notifications. The mobile allows you to receive notifications, but instead of traditional messages in popup format, the Callback 8020 makes use of a small LED screen on the front cover that displays characters with a very retro design reminiscent of the old calculators from the 70s and 80s. Eighties sounds. Audio includes a music player for the 8-bit SID chip, which was used in the original C64. It also has, of course, an integrated DAC, an F radio and a 3.5 mm jack. There are even in-ear headphones included in the box. Shell format, interchangeable shells. The design of the mobile phone itself is most peculiar: it adopts a clamshell format, which makes the interior screen relatively small in size, and it has a physical keyboard and a small directional pad on top of that keyboard. The outer shell can be exchanged. Commodore Callback 8020, in its Founders Edition variant, at $640. Source: Commodore. lor retro is not at all cheap. The price ranges between $500 and $640 depending on the variant chosen — the most expensive is the Founders Edition, with gold tints. It is a high price if one only looks at the specifications and design, but that is not what you pay with this mobile. What you pay is more for the nod to nostalgia and for the system that tries to prevent us from being glued to the screen. We are faced with a different concept, which may be shocking and strange, but which tries to solve a problem and also do so with an original approach. I don’t know if the original Commodore would have made a mobile like that, but one thing is certain: the launch is, to say the least, original, loud and brave. That’s something. In Xataka | While the Spectrum and Commodore 64 were triumphing around the world, there was a terrible problem in Japan: kanji

I have tried all the browsers with AI on the market and the one that works best is the one that started the worst: Dia

by

I started using Arc in the summer of 2024. I was slow to get into it (I had tried it before and found it too capricious, too determined to teach me how to sail when I already knew) but when I arrived, I really arrived. Vertical tabs. The spaces. The way he organized the digital day without me having to think about it too much. Arc to me was not just a browser: it was a work environment. And then The Browser Company Dia announced. I remember it with some annoyance. Josh Miller, the company’s CEO, uploaded a video talking about “something new,” and Arc was in “minimal maintenance,” a euphemism for not saying the word “abandonment.” The Arc community complained, rightly so, and I joined the discontent. They had built something great and threw it overboard at the first opportunity to pursue the AI ​​chimera. Still, I tried Dia as soon as I could and wrote about it: You opened Dia and saw Chrome. Chrome with better typography, Chrome with more careful animations, but Chrome nonetheless. No split tabs, no gaps, no anything that had made Arc great. Just a chatbot in the sidebar. I closed it and went back to Arc, which continued to work despite the abandonment warnings, until I started trying other AI-based browsers. Months passed and Dia was updated, week by week, with a striking cadence. And at some point I started to notice something: the things I missed were coming backsometimes even improved. First the vertical tabs, which Arc had popularized and which Chrome has just announced that it will also adopt, something that says a lot about who sets the pace in browser design. Then the groups of eyelashes, with that aesthetic care that has always been a trademark of the house and now goes further than before. Other browsers already have this feature, but “not like this.” Recently, the split viewwhich I have been using for years in Arc and which is one of those functions that, once you have it, you don’t understand how you navigated without it: Simultaneous view of several tabs in a single window. Split tab view to display multiple tabs in a single window. Image: Xataka. Tab groups have even partly replaced my use of favorites. You can create them by hand or see how they are automatically grouped and renamed when you open multiple tabs from the same place. Its design and user experience are fantastic. Image: Xataka. The set favorites are still there. Tabs in the sidebar also ended up coming to Dia. And between both blocks, the groups of tabs. Image: Xataka. There is a pattern in The Browser Company that you should have already learned with Arc: they release something that seems incomplete, almost “psché”, and then they improve it until you can’t put it down anymore. It took me two tries to fall in love with Arc. Something similar has happened with Dia, only the process has been longer and the reconciliation more gradual. And that boss now has to live with a new owner, because since September This company was bought by Atlassianwhich wants to make Dia the reference for working with AI. Nothing has changed at the moment on a day-to-day basis (pun intended) of the browser. What Dia has done smartly (and differently than Arc) is start almost from scratch and give up all the weirdness. Arc asked you to adapt to it, to learn its logic, to assume its curve. And Dia does exactly the opposite: she is very Chromiumvery familiar, very little foreign to anyone coming from Chrome and not as rigid as Arc. That takes a toll: some of Arc’s more radical ideas have been lost in the transition, but it also means it doesn’t create as much friction for the average user. You open Dia and browse. There is no twenty-minute tutorial on how to think about tabs, which is something that penalized the growth of Arc: those of us who used it loved it, but many people left when they saw that they did not understand the proposal. The weight is what catches my attention the most on the negative side: it’s around gigabyte, which is an outrage for a browser. And there is no mobile version yet. That hurts more, because it means the experience is split between devices, and the consistency Arc was trying to offer across platforms doesn’t yet exist on Dia. I hope they don’t take long. In theory it will arrive this year. Regarding AI, it is not what I use the most. Of course I use Claude or Gemini, I mean Dia’s chatbot sidebar. It’s not something I used on Dia nor have I used much on Neon, Atlas or Comet, the other three that I’ve tested in some depth. The chat in the sidebar is a good complement, I can ask it something about the tab I have open, ask it to summarize several at once, respond with context of what I am reading; but it is not the focus of my browsing experience. All three browsers conveyed at their launch that AI would change everything, and reality is more modest– It’s useful, sometimes very useful, but it doesn’t transform your workflow in the same way that a good tab design does. At least in my experience. I have tested more browsers that leverage AI for their value proposition. And they all have their strong points: Comet It is very fast and efficient searching for information in real time. Atlas is very capable when you need to systematically extract data from multiple pages, but none have the level of care in the experience that Dia has. Opera Neon is a browser built from the ground up for the AI ​​era, with much more than a well-placed chatbot. But it’s not just that they work. Dia feels good. There is something in the design, in the animations, in how the groups of tabs are constructed, that remains … Read more

Browsers prepare for the most radical transformation in their history. One in which the IA will be

by

Do you remember what was the first Web browser What did you use? Whatever your answer, the truth is that, essentially, the way to use them has not changed too much since Netscape Navigator. Yes, over the years, browsers have lived deep transformations. At the beginning they endured flat pages, almost like a digital newspaper, and today they carry websites full of photos and videos on their backs. It was they who brought us the eyelashes, who changed forever the way of organizing what we had open. Security was also consolidated with the lock of the HTTPSthe responsive design that allowed the web to the mobile without losing usability and more. All this was marking important stages, but the truth is that, essentially, we never stopped Enter, write, click and handle the pages ourselves. Advances there were many, but the interaction remained the same That dynamic, however, could be about to transform. And there are weight reasons to think that it will be. The technology industry has focused on automation, in a scenario where AI agents They assume a good part of heavy work. These systems aim to be capable of: Plan and divide complex tasks into logical steps. Choose the best tools to meet what they are asked. Have memory and context to meet the user and offer more tight solutions. Imagine agents capable of managing tasks from beginning to end. From Organize a trip With hotels and flight reservations at the best price, until the weekly purchase, create spreadsheets or handle specialized software. It would be enough to give them instructions (and the appropriate permits) to act in our name and consult us only when necessary: ​​confirm a purchase, choose an option or solve a problem in the process. The interaction with them would be direct: text in a chat window or, better yet, the voice. Nothing to fight again with menus, forms or traditional interfaces of pages and web programs. The truth is that we are already seeing steps in that direction. Perplexity has Cometan AI browser that Execute tasks on the username. Openai presented Operatorthen integrated into the agent mode Chatgpt With your own browser. Claude’s extension for Chrome in Action Anthropic, meanwhile, has entered the race with An extension of Claude capable of controlling Chromewhile Google develops PROJECT ASTRAan agetic system that goes beyond the browser and can operate directly on Android devices. For now, most of these proposals are in an experimental phase and with limited deployment. But it would not be strange that, as quickly as the sector progresses, they soon become a new way of interacting with the web. Comet, the agentic browser of Perplexity Just remember how we worked before November 30, 2022, when Chatgpt did not exist yet. Barely two years and eight months latertechnology has radically changed the panorama. Of course, not everything is promises. One of the biggest challenges of browser agents is security. And here the edges are many. Nobody wants an agent who, for misunderstanding instructions, reserves a wrong flight or buy something unnecessary. Neither do we want them to become Victims of cybercriminals, that they are already exploring ways to manipulate them just like before they tried to deceive us with phishing or scams. Malicious instructions are a clear example: if a human can be deceived, you can also try to make an agent execute actions against us, such as revealing passwords or extracting sensitive information. The question is inevitable:We are entering the era of artificial intelligence agents in the browser? Would you use them? Images | Xataka with Gemini 2.5 In Xataka | If you don’t know why your Instagram has been filled with orcs and sharks, the answer is simple: a false video made with ia

AI has just taught us the future of the most important application of our life: subscription browsers

by

How much of what you do on the computer do it in the browser? Exact. A lot. We don’t think too much about it, but The browser is the application by antonomasia of our time. Not so much on the mobile, where native apps – which are often browsers disguised as something else – triumph, but of course in desktop and laptops. The Internet has been the indirect guilty of this transformation of the browser in one of the pillars of our life. The greats of technology know it well, and for more than three decades they have tried to conquer with their browsers to dominate the world. Of that mythical Netscape We become absolutely dominated by Microsoft Internet Explorerand after the occasional scarce with Firefox —Now in really low hours– We end up yielding to a Wonderful and disturbing Google Chrome. This company, by the way, was clear from the beginning how important the browser was. So much that they created an operating system absolutely focused on that idea: they called it Chrome Os. During all this time, yes, there was something inherent to all those browsers: They were free. Microsoft, Apple, Mozilla, Opera and of course Google gave them to us. They did not need to charge for them because they earned (and earned) a lot of money indirectly with their integration of search engines, especially Google’s. They tell Apple: Google pays about 20,000 million a year so that your search engine is the native in Safari. If you pay so much, imagine what you earn through the searches made by users of the iPhone, iPad or Mac. Thus, using the browser has always been something we have done without thinking that it could cost us money. That will change very soon. Hello, subscription browser And it will change because before us a new era opens in which browsers want to be much more than that. And to achieve this they have allied with AI models that allow interaction with websites to be much more powerful. The promise of browsers with AI It is exceptional. One of the first, Perplexity Cometit autocalifies as “the browser that thinks with you.” Like him, browsers with AI are able to summarize that article that you are reading or answer questions about it, integrate conversational engines and little by little they are beginning to integrate something especially promising: AI agents What do things for us. That click alone and fill for forms and that they even overcome captchas without help. But do that this time will not be free. Not at least if we want to use those options intensively. This is what The Browser Company is already raising, the company that first developed Arc and? He abandoned it To launch your own browser with the call day. Josh Miller, CEO of The Browser Companyhe confessed in An interview With The New York Times that in the coming weeks will launch subscription plans that will go from five dollars a month to hundreds of dollars a month, depending on how much users want to take advantage of the NA of the browser. There will still be a free version that will allow a limited use (probably very) of these options, but the future of this browser is not to be a browser, but to be a subscription browser. The decision made by this company is totally logical: Using AI is not cheap, and using it is still less cheapso if you want to do it, you will have to pay. It is the same as we are seeing with the different models: Chatgpt, Gemini or Claude can be used for free, but if you want to use them intensively and access their most advanced options, it is time to pay by pointing to their subscription plans, as is the case with Chatgpt Plus, Google Ai Pro or Claude Pro respectively. Browsers are already the most widespread way to interact with AI, so it is logical to think that all of them – Hello, Firefox? – will end up integrating these functions. He has done it day, he has done it, he just did it Edge with your co -pilot mode And it is expected that Integration between Gemini and Chrome So it is much more direct. All of them will remain free in their basic versions. We can normally navigate as we have always done, but if we want to go further, if we want to use all those attractive AI options, Touch to pay a subscription. One that will be increasingly expensive. We will have to go to the idea. In Xataka | Anthropic has seen that their users do not stop using the 200 euros plan a month of their AI. They had to stop their feet

Perplexity is going to launch its own browser. It is one more proof that traditional browsers fall short in the AI ​​era

by

Perplexity He has announced Cometa browser specifically designed for the AI ​​era. It is part of something else: of the trend in which several technology seek to reimagine how we interact with the web. Why is it important. The next wave of A -centered browsers reveals an uncomfortable truth: traditional browsers are not optimized for a world in which AI agents will act as intermediaries between information and us. Chrome, Safari, Edge or Firefox can improve whatever they want in these capabilities, but IA deserves a creation from scratch. The Browser Company understood and paralyzed the development of ARC to create something new: day. The context. Perpleplexity has opened its waiting list for Comet, its next browser that promises to “reinvent” our way of accessing the web. This movement comes just weeks after OpenAi will present its operator agentand in the middle of Rumors about a possible OpenAi’s own browser. Current browsers were designed for humans who use keyboards and mice, not for ia agents who need to interact with the website of radically different shapes. Yes, but. The first generation of AI agents as Openai Operator uses chrome modified versions to complete tasks, offering an almost theatrical show: we see a bot by moving a cursor and writing how a human would do. It is effective, but also inefficient. Chrome use as a base poses another problem: what will happen when Google completely integrates Gemini In your browser? Rivals such as Openai and Perplexity will be vulnerable to Alphabet. New era. The race to create the final browser for the AI ​​era remembers the Navigators War of the 90s, when Netscape and Microsoft struggled to dominate Internet access. That conflict initially won Microsoft with Internet Explorer, but Google Chrome, which did not even exist then, ended up dominating the market years later. The lesson is clear: the winner of this new battle can be a company that still does not have a browser. Or that it has not even been founded yet. Several actors are already on stage: Perplexity with Comet, the aforementioned The Browser Company with Dia, and rumors on similar OpenAi projects. The integration of AI models such as Claude, Gemini and GPT directly in the navigation experience will be key. Between the lines. True innovation does not consist of adding functions from AI to existing browsers, but completely reimagining the navigator concept. A browser designed for the AI ​​era should allow both traditional human use and the efficient functioning of autonomous agents, perhaps with different ways according to who is “behind the wheel.” The browsers Ai-first They will probably dispense with interface elements designed for humans when they operate in automatic mode, using more efficient APIS and communication channels to interact with the web. A good simile is that of the autonomous car: if it is able to drive on its own, it makes no sense that it makes a steering wheel, pedals and five seats looking forward: you can and must reimagine the cabin completely. Deepen. For companies such as Perplexity, launching its own browser not only involves diversifying its product offer, but also ensuring their independence from the platforms controlled by their competitors. Same incentives that OpenAI has to launch its own proposal. The browser has historically been the entrance door to the Internet, and whoever controls this door will have a disproportionate influence on our digital experience. Google knows it well: Chrome has provided you a huge strategic advantage. We are witnessing the birth of a new category of software that could transform our relationship with the Internet as much as the original browsers did. The question is not whether browsers Ai-first They will replace the traditional ones, but when they will do it and who will lead this transformation. Openai looks like the great threat to him status quo from Google. And there will be more. Outstanding image | Perplexity, Xataka with mockuuups studio In Xataka | Mozilla’s long crisis: an eternal users and a Google dependency that is still majority

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.