Latest news

Microsoft's GitHub, the world's largest code repository for software development projects, will start using its users' program codes as training material for its own AI.

The change begins on April 24, 2026, and it does not apply to regular GitHub usage, but only to those users who also use GitHub's own GitHub Copilot.

After the change, all conversations had with Copilot, prompts, and code generated by Copilot will be taken as part of Copilot's AI training material. According to Microsoft's statement, the change will significantly improve the quality of code generated by GitHub Copilot, as it will henceforth be based on real life and conversations developers have already had with the AI.

What makes the change unfortunate is that it is a so-called opt-out solution: all GitHub Copilot users are automatically subject to the new policy, unless they themselves refuse to share their activities as AI training material.

The explosive growth in the prices of memory chips and components, which began in autumn 2025, is gradually extending to new product categories.

Now, Japanese Sony has announced (in Japanese) that it is suspending the sale of memory cards for the time being.

According to the company, the background is supply difficulties, meaning Sony is unable to acquire enough NAND memory chips, which are used as storage media in SSDs, memory cards, and USB flash drives.

Data centers rapidly being built around the world for the needs of AI companies are consuming almost the entire market for both operating memory (RAM) and NAND memory - and in consumer products, the situation is reflected in both rising prices and availability problems.

Sony's announcement of the sales suspension concerns both SD and CFexpress memory cards.

Already yesterday, Sony announced another change, which is also linked to the explosive growth in component prices:

Austria has announced its plan to introduce a legislative change that would prohibit the use of social media by children under 14 years of age.

Austria's initiative is part of a widespread populist trend where several countries are considering restricting children's online use to protect young people from the harmful effects of social media. According to BBC information, the Austrian government has already held extensive negotiations on the matter, but details regarding the implementation and timeline of the law have not yet been confirmed. According to the government, the framework of the legislative initiative could be ready as early as June 2026, after which it would move to a more detailed evaluation process.

The first country to categorically ban all social media services for minors is Australia, where social media was banned for under 16s just before the start of Australian schoolchildren's summer holidays, in December 2025. Similar restrictions have either come into force or are planned elsewhere to some extent: in France, the age limit for social media services is 15 years, and Spain and Ireland are working on restrictions for under 15s.

Several serious security vulnerabilities have been found in many routers from the hugely popular Chinese company TP-Link.

The vulnerabilities have been found in these models, widely used even in Finland:

• TP-Link Archer NX200
• TP-Link Archer NX210
• TP-Link Archer NX500
• TP-Link Archer NX600

The most serious of the discovered vulnerabilities is CVE-2025-15517, which allows for the hijacking of router control without any user interaction. The vulnerability has a risk level of 8.6/10 and is related to a bug in the router's browser-based management panel, which uses an http connection, allowing login to be bypassed and access to the router's administration views.

Other discovered vulnerabilities also have a risk level of 8.5/10: CVE-2025-15518, CVE-2025-15519, and CVE-2025-15605.

The company has released an update for all of these, in the form of new firmware. Unfortunately, routers typically do not update themselves, so every router owner should handle the update manually.

Google has published a new, significantly faster - and more worrysome - timeline for the so-called encryption apocalypse. According to the company, current forms of encryption technologies will become insecure as early as 2029.

This is the so-called Q-Day (quantum day), described as the moment when the computational power of quantum computers reaches a level where public-key encryption algorithms (such as RSA) and elliptic curve-based encryption algorithms can be broken.

Previously, encryption professionals estimated that the moment would not arrive until the 2040s, but now Google estimates that this moment will arrive in just a few years. According to Google, the faster timeline is due to the rapid development of quantum computing hardware recently, and also the swift progress in solving quantum computing-related problems, such as error correction.

The situation is serious, as hostile states and criminals have very likely been collecting encrypted communications, messages, emails, and encrypted documents for years. They are just waiting in some corner of a hard drive for the moment when quantum computing reaches that dreaded "encryption apocalypse" - and then all previously collected secrets can be opened quite easily. This practice is called the Harvest now, Decrypt later method in cybersecurity circles.

OpenAI has announced that it is discontinuing its Sora service, which creates videos using AI and had become quite popular.

Sora was opened for public use only a few months ago, but it has garnered criticism from copyright owners who have accused it of stealing their characters for AI model training material. At the same time, OpenAI has increasingly found itself as an underdog in the AI race, where Google's Gemini is growing rapidly and another competitor, Anthropic, is luring developers to its Claude Code.

So for OpenAI, it's partly about concentrating resources on the company's most important products - which Sora clearly was not among. The company confirmed Sora's discontinuation in its social media post (link leads to X/Twitter).

At the same time, media giant Disney has decided to discontinue the cooperation with OpenAI announced in December. The purpose of the cooperation was specifically to focus on using the technologies behind Sora in the production of Disney's films and series.

FCC shook the technology markets this week by completely banning consumer-grade routers manufactured outside the United States.

The reasoning behind the ban is that the foreign routers pose a threat to United States' national security. American lawmakers have been suspicions about the data security of Chinese products for quite some time - and the ban reflects those fears. But instead of the FCC banning only routers made in China, it simply banned all routers manufactured outside the United States - whether they are manufactured in China, Finland, or Canada.

So, no exception is made even for routers manufactured by American companies if they are manufactured somewhere other than the United States. For example, all Google's Nest routers are manufactured outside the United States, and therefore they too will be banned in the future. The same ban also applies to American network device giant Netgear, whose products are not manufactured in China, but also not in the United States (but e.g. in Vietnam).

Mozilla has announced big updates to Firefox, which improve user security and privacy.

Perhaps the most interesting of the announced updates is a free, built-in VPN. All Firefox users are promised 50 gigabytes of free VPN usage monthly going forward. According to Mozilla itself, free VPN services on the market are usually at least questionable regarding privacy protection - and Firefox's built-in VPN offers a solution precisely to this problem.

The VPN will be available already by the very end of March 2026, but only in the United States, France, Germany, and Britain. No timeline estimates were given yet as to when the VPN service will expand to other countries.

Firefox also takes a significant step forward, as it becomes the first browser to support the new Sanitizer API. This is a web security standard that helps prevent cross-site scripting (XSS) attacks, which are one of the most common and harmful web vulnerabilities. With the Sanitizer API, developers can easily clean untrusted HTML content before adding it to a webpage, which reduces vulnerabilities without requiring major changes to existing websites.

Microsoft has announced significant changes to its Windows 11 operating system, particularly by reducing the visibility of the AI assistant Copilot in various parts of the operating system and its applications. The decision is part of a broader strategy aimed at improving Windows' quality, performance, and user experience.

The changes are a direct response to user feedback, which indicated that Copilot's integration has at times been excessive and even disruptive.For the past few years, Microsoft has focused on pushing Copilot into virtually every possible corner of Windows, starting from default Windows applications like Notepad. User reactions to these AI features have been mixed: while some appreciate the possibilities brought by AI, many have found Copilot to be too prominently displayed and difficult to remove.

Criticism has also arisen from the fact that AI is not always genuinely useful in all applications, and its addition seemed primarily to bloat the operating system. Microsoft's corporate vice president for Windows and Devices, Pavan Davuluri, stated in the company's blog that Copilot will no longer be "shoved everywhere" in the future. Instead, AI functionalities will only be added to parts of Windows where they are genuinely believed to bring benefit. Initially, these changes will be visible in applications such as Windows' Photos app, Notepad, and the Snipping Tool, where its visibility will be significantly reduced.

AI bots will surpass human users on the internet by 2027, estimates Matthew Prince, CEO of web infrastructure company Cloudflare.

According to Prince, a historic change is occurring in web traffic, driven particularly by the rapid proliferation of generative AI - such as ChatGPT and Google's Gemini. This change affects both online services and who will use the internet in the future, and how.

Matthew Prince spoke about the phenomenon at the SXSW conference in Austin, Texas. According to him, about a fifth of internet traffic was long generated by bots, but with the advent of generative AI, the proportion of bots has begun to grow sharply. One reason for this is the operating method of AI agents: when a human shops or plans a trip, they visit a few websites, but an AI bot can easily search for information from thousands of sites in the same amount of time. As a result, a single AI agent can generate many times more web traffic compared to a human.

AI bots are also not limited to search engines or simple automated tasks. New generation AI agents perform complex tasks independently on behalf of users - comparing products, planning holidays, and compiling information from multiple sources. This significantly increases so-called agent-based traffic, which burdens the internet's underlying infrastructure, servers, and security services.

Google is bringing significant changes to installing Android applications from sources other than Google's official Play Store. According to the advertising giant, the goal is to improve user security and combat malware - but at the same time, the lives of technically savvy users will become more difficult.

The changes specifically concern situations where users install applications from alternative sources, and the applications are not available for download through official app stores. Previously, Android's openness allowed applications to be installed freely, but at the same time, it also exposed users to malware.

App developers will henceforth be required to verify their identity when registering as app distributors, which is expected to reduce the spread of scam and malware. The developer must provide Google with their name, address, email, and phone number, among other things, and pay a one-time fee of 25 dollars (approximately 20 euros).

Starting in September 2026, Google will gradually introduce a new so-called advanced flow practice, which specifically concerns more technically proficient users. If a user absolutely wants an application on their phone or tablet that has not been verified by Google, the installation must go through a new, rather complex-looking process. The process includes several steps designed to ensure that the user understands the risks - and is not, for example, acting on the instructions of a scammer.

A significant crack has appeared in the years-long, close collaboration between Microsoft and OpenAI in recent weeks, as information about a potential legal dispute between the companies has emerged.

According to sources (paywall) from the financial newspaper Financial Times, Microsoft is considering suing OpenAI for breach of contract, stemming from OpenAI's recent collaboration agreement with Amazon's cloud service, AWS. The agreement with Amazon may violate the exclusivity agreement between Microsoft and OpenAI, which stipulates that certain AI models developed by OpenAI should only be available through Microsoft's Azure cloud platform.

The partnership between Microsoft and OpenAI has been exceptionally close since 2019. Microsoft has invested over $13 billion in OpenAI over the years and, through the deal, acquired approximately a 27 percent ownership stake in the AI company. The agreement between the companies stipulates that OpenAI can collaborate with other cloud service providers, but this activity has strict limits. Specifically, the enterprise platform named Frontier, designed for developing and managing AI agents, has been agreed to operate exclusively on Azure.

Streaming service Spotify has released a new feature aimed at users particular about sound quality - so-called audiophiles. The new exclusive mode offers the ability to play music on a computer as purely and bit-perfectly as possible, without the computer's own audio processing altering the original sound. The feature is available for Windows and will also be coming to Macs later.

Exclusive Mode is enabled from the settings of the Spotify desktop application. When the mode is activated, Spotify takes control of the computer's audio output - for example, via an external DAC converter - and completely bypasses the operating system's own audio processing. By completely bypassing the computer's operating system audio processing, it is possible to avoid loss of sound quality and the mixing of the operating system's own sounds with the music. As a result, the user hears the music as close as possible to how it sounded in its original form.

Enabling the feature requires the user to have an external DAC or other audio interface - in practice, many computers' built-in headphone or speaker ports are also compatible with exclusive mode.

The world's most famous printed dictionary, Encyclopedia Britannica, and its subsidiary Merriam-Webster have sued AI company OpenAI, accusing the company of copyright and trademark infringement.

Encyclopedia Britannica alleges that OpenAI has used the content of its encyclopedia and dictionary to train its AI models without permission. According to the lawsuit, OpenAI allegedly copied nearly 100,000 articles from Encyclopedia Britannica - and due to this copying, OpenAI's ChatGPT is able to reproduce Encyclopedia Britannica's content almost verbatim.

According to the publisher, the AI-generated responses are not limited to the training phase; ChatGPT is sometimes able to produce partial or complete copies of protected articles. Britannica considers this a threat to its website traffic and business, as users can obtain information directly from the AI without needing to visit the original source.

Britannica also raises concerns related to trademark infringement. According to the company, ChatGPT may in some cases include references to Encyclopedia Britannica in its responses in a way that could give the impression that OpenAI is using their material with permission. Furthermore, the AI can produce incorrect or fabricated content that is erroneously attributed to Encyclopedia Britannica. The company states that AI-invented claims, combined with the Encyclopedia Britannica brand, damage its reputation and credibility built over more than two hundred years.

Garmin and Meta have started a new collaboration, making WhatsApp now available on Garmin smartwatches. Users can now read and reply to their WhatsApp messages directly from their wrist.

Previously, Garmin watch users could only receive notifications of incoming messages, but replying was not possible, especially with devices connected to an iPhone. With the new WhatsApp application, users can now read chat history, reply to messages using the built-in keyboard, and use emoji reactions. Additionally, the app allows users to see incoming calls via WhatsApp and reject them through the watch. Messages remain under WhatsApp's end-to-end encryption, so privacy is maintained even when used separately from a mobile device.

WhatsApp is the first instant messaging application released for Garmin watches, and its availability is currently limited to certain models. These include Garmin Forerunner 970, Fenix 8, Venu 4, and Forerunner 570. Older models and slightly more specialized models, such as the Instinct series or golf-oriented Approach models, are currently not supported. The application also requires the user's phone to be connected to the watch and WhatsApp to be installed on the phone as well.