Tech
Software, devices and AI, with a short English summary and a link to the original article. RSS
-
Tech
Xbox embraces AI-assisted ports of classic games
Microsoft announced an Xbox initiative around decompiling and recompiling older games. These techniques, often helped by AI, let fans bring classic titles to new hardware. Instead of fighting them, the company chose to make use of them. The move could change how video game history is preserved.
What it means: Microsoft choosing to embrace the decompiling and recompiling of old games, rather than fighting it in court, is a notable shift for an industry that has traditionally guarded its code tightly, even for titles no longer on sale. These techniques, increasingly helped by AI, let groups of fans reconstruct a game's source code from its executable and make it run on modern hardware, which would otherwise take an enormous amount of work. For preserving video game history the benefit is large, since many classic titles risk disappearing once the hardware they were built for stops working. The same technology matters beyond games too: many businesses still run old systems with no documentation left and no original developer to ask, and AI that helps understand and port legacy code could make modernising them much cheaper, provided it is done with proper review and testing before anything goes into production and with respect for licences.
-
Tech
SpaceXSI bundles X Premium, Grok and Cursor in one plan
The new X subscription combines the platform's paid features with AI tools such as Grok and Cursor. Plans range from $8 to $200 per month. The aim is to make the subscription more appealing to professionals and developers. It also reflects a trend of packaging AI services with social networks.
What it means: The new SpaceXSI subscription, which combines X's paid features with AI tools such as Grok and Cursor, follows a familiar logic: when one product on its own is not enough to make people pay, bundling several services makes the price feel more reasonable, as TV and telecom providers did in the past. The range from 8 to 200 dollars a month shows the bundle targets ordinary users as well as professionals and developers who use coding tools every day and will pay more for higher limits. For someone already paying separately for AI tools, the bundle may lower costs, but it is worth checking usage limits first, what exactly each tier includes and whether all the services are really needed. Tying productivity tools to a social network also raises data questions, because when one company runs more of a user's services it gathers more information about them, which businesses should weigh when choosing tools for their team.
-
Tech
Siri on iOS 27 learns your preferred music app
Siri can now learn which music app a user prefers. Spotify users can ask for songs, podcasts and playlists without adding 'on Spotify'. It is a small but long requested change in Apple's ecosystem. Many link it to regulatory pressure for more open platforms.
What it means: Siri on iOS 27 learning which music app a user prefers looks like a small change, but for years it was one of the clearest examples of Apple favouring its own services: Spotify listeners had to name the app every time, while Apple Music was the automatic choice. The change comes during intense regulatory pressure, especially in Europe under the Digital Markets Act, which requires large gatekeepers not to favour their own products over competitors. For users the benefit is practical and immediate, with less friction in daily use, while for app makers it signals that integrating with the voice assistant and system features is worth more, since third-party apps can now become the default. The broader message is that platforms open up gradually, usually not voluntarily but under the pressure of rules, and anyone designing mobile products should track which new capabilities each release makes available to them.
-
Tech
Pluto Security raises $42M to tackle workplace AI risk
Pluto Security has raised $42 million in funding. The company focuses on the problem created when every employee gets access to AI tools and uses them in unpredictable ways. Security teams worry this opens new gaps in their systems. The round shows that AI security is becoming its own market.
What it means: Pluto Security's 42 million dollar round confirms that securing AI inside companies has become its own market, with a distinct problem to solve: when every employee has access to chatbots, coding assistants and agents, company data can end up in outside services without anyone knowing, and automations can gain permissions nobody explicitly approved. Traditional security teams were set up to control applications and networks with predictable behaviour, whereas AI tools are used in unpredictable ways, often by staff who simply want to work faster. For smaller companies without a security team, the answer does not have to be an expensive platform: a clear policy on which tools are allowed, which data never goes into them, and company accounts instead of personal ones already provides visibility. Visibility into usage is the first step, because you cannot protect something you do not know exists, and GDPR requires you to know where personal data goes in any case.
-
Tech
FBI arrests an OpenAI engineer over a hidden recording device
The FBI arrested an OpenAI engineer after the company found a sophisticated recording device hidden in an office meeting room. The 35-year-old is charged with intercepting or attempting to intercept spoken conversations. According to the FBI director, the aim was to gather sensitive details about the company's superintelligence work. The case shows how valuable AI lab secrets have become.
What it means: The arrest of an OpenAI engineer over a hidden recording device in a meeting room is a reminder that the most valuable secrets are not always stolen over the internet: they can leak through someone with legitimate access to the building who knows where and when important conversations happen. The FBI director linking the case to information about the company's superintelligence work shows that authorities now treat AI labs almost as strategic infrastructure, with a matching interest in espionage. It should be noted that this is a charge and not a conviction, and the motives and possible recipients of the information have not been made public. For any organisation with sensitive projects, the practical takeaway is that security needs physical measures as well, such as room sweeps, restricted access to certain areas and a device policy, not just firewalls and encryption, while managing insider risk means balancing oversight with trust in employees.
-
Tech
Court orders OpenAI to share hiring records with Apple
A federal judge ruled that Apple may access OpenAI's interview material with former Apple employees. The order also covers private messages and depositions from key people in the case. Apple accuses OpenAI of misappropriating trade secrets through its hiring. The ruling adds pressure on OpenAI as the case moves forward.
What it means: The judge allowing Apple to access OpenAI's interview material with former Apple employees, along with private messages and depositions, significantly raises the stakes for OpenAI, because that evidence could show exactly what was discussed during hiring and whether confidential knowledge changed hands. Competition for talent in artificial intelligence and hardware is so intense that engineers moving between large companies are increasingly followed by legal disputes, and each case shapes practice across the sector. It should be stressed that the ruling concerns access to evidence and not the substance of the allegations, which have not been decided. For any company hiring from competitors, even at a far smaller scale, the lesson is practical: it needs a documented process showing that a new employee brings no documents, code or information from a previous employer, together with clear confidentiality terms from the very first day, so that hiring good people does not turn into an expensive legal dispute later.
-
Tech
Weighing the benefits and risks of open-weight AI models
MIT Technology Review examines the debate over AI models with open weights. It was prompted by an Anthropic report that called the Chinese model GLM-5.3 a serious cyberattack risk. On the other side, open models offer transparency, lower costs and independence from a few vendors. The piece weighs both sides without an easy answer.
What it means: The debate over open-weight models is no longer theoretical, and the trigger for this piece, an Anthropic report calling GLM-5.3 a serious cyberattack risk, shows how fast the gap between open and closed models is closing on capabilities that can also be misused. On the other side, the benefits are concrete: a business can run an open model on its own servers, keep its data inside the EU, inspect how it behaves and avoid depending on a single vendor's prices and terms. It is also worth noting that a risk assessment of a model by a competing company inevitably has a commercial dimension, so reports like this need independent confirmation. For small and medium businesses the practical question is technical rather than ideological, because choosing an open model means taking responsibility for secure deployment, updates, access control and filtering, whereas a hosted provider handles part of that in exchange for less control.
-
Tech
Regulators review reports of Tesla robotaxis speeding in school zones
The US road safety agency NHTSA is looking into reports that Tesla's roughly 1,000 robotaxis exceed speed limits in active school zones. The review covers the company's self-driving system. Similar probes have targeted Waymo. The outcome will affect how quickly autonomous taxis expand.
What it means: NHTSA looking into reports that Tesla's robotaxis speed through active school zones touches the most sensitive point of autonomous driving: not routine motorway journeys, but temporary local conditions, such as a limit that applies only when pupils arrive and leave and is signalled by flashing lights or changing signs. For a system built on cameras and maps, interpreting rules like these correctly is hard, and that is exactly where readiness for wide driverless use gets decided. Similar probes into Waymo show the problem is not unique to one company but typical of an industry that is still learning, while Tesla's fleet of around a thousand vehicles makes the issues more visible. For regulators, the case will influence how quickly robotaxis are allowed to expand into new cities, and for the public it is a reminder that trust in such systems builds slowly and could vanish after one serious incident near a school.
-
Tech
Mind Robotics hires former Ford EV chief Doug Field as president
Mind Robotics, the robotics company led by Rivian chief RJ Scaringe, has hired Doug Field as president. Field ran Ford's electric vehicle work until recently. Earlier he led Apple's Project Titan and the development of the Tesla Model 3. The hire is seen as a major win for the secretive venture.
What it means: Mind Robotics hiring Doug Field shows where talent and capital are heading in technology: an executive with experience on the Tesla Model 3, Apple's Project Titan and, more recently, Ford's electric vehicles chooses a robotics company connected to Rivian's chief, suggesting robotics is seen as the next big frontier after electric cars. Field's background matters for a specific reason, because it is about mass production of complex hardware, exactly where many robotics companies fail, since a prototype that works in the lab is a long way from thousands of units with consistent quality and sensible cost. Mind Robotics remaining largely secretive makes it hard to judge what it is building, but hiring a profile like this signals industrial-scale ambitions. For the wider sector, executives moving from carmaking into robotics indicate that automotive skills such as supply chain management, safety engineering and quality control are becoming valuable in new fields.
-
Tech
GTA 6 reportedly heads to PC in early 2027
Reports say the PC version of GTA 6 will launch in early 2027. If true, the wait will be far shorter than for GTA 5, which took years to reach PC. Rockstar has not announced anything officially. For PC players it is one of the most anticipated pieces of news this year.
What it means: Reports of a GTA 6 PC release in early 2027 matter beyond gamers, because they show how release strategy is changing for big titles: publishers used to hold the PC version back for years, partly for technical reasons and partly to sell on consoles first, but today the PC market is too large to leave out for that long. If the report is confirmed, the contrast with GTA 5, which arrived years later, will be striking and will show that development tools now make it easier to move between platforms. It should be stressed that Rockstar has announced nothing officially, so the date remains a rumour and could change, as often happens with titles this large. For the industry, a launch of this size also affects the hardware market, since many players upgrade graphics cards to play, and technology retailers usually prepare for higher demand around such dates, both online and in physical stores.
-
Tech
Google Play now hosts private AI-built Android apps
Anyone with a Google account can now publish private Android apps built with AI in Google AI Studio to the Play Store. These apps are not listed publicly, run only on the creator's own device and need no paid developer account. They must still follow Play's rules. The feature starts with adult users in the US, alongside new AI tools for finding apps and games.
What it means: Letting anyone publish private AI-built Android apps without a paid developer account changes who can make software for personal use: a professional can describe a tool for their own needs in plain words, such as a cost calculator or a task list, and have it on their phone within minutes. Because these apps run only on the creator's device and are not listed publicly, risks to other users stay limited, while the requirement to follow Play's rules shows Google wants to keep control over anything that passes through its store. For the software industry, this does not make professional developers redundant, but it shifts their value towards what a description cannot solve: security, integration with other systems, scaling to many users and long-term maintenance. The initial rollout only to adults in the US suggests Google is moving cautiously, and the feature may reach Europe once the risks and regulatory requirements have been assessed.
-
Tech
Google rolls out Gemini 4 Argon, for paying users only
Google has started the general release of Gemini 4 Argon, its most advanced model, a few weeks after announcing it. Developers can use it through the API, and Pro and Ultra subscribers get it in the app, with an introductory half price until late November. Google highlights complex coding and autonomous tasks, with a limit of up to one million tokens. A new safeguard halts runs when it detects a suspicious workflow.
What it means: Making Gemini 4 Argon generally available only in paid plans and through the API shows Google treating its strongest models as a high-value commercial product rather than a free way to attract users, which fits their high running costs. The focus on complex coding, autonomous tasks and a limit of up to one million tokens means the model is aimed at teams that want to hand it entire projects, large document sets or big code repositories, not just short questions. The mechanism that halts a run when it detects a suspicious workflow is an important detail, because it shows vendors taking on some responsibility for what agents do, but also means a legitimate job may occasionally stop without warning, which has to be planned for in system design. For businesses evaluating the model, the half-price introductory period until late November is a good window for testing on real scenarios, measuring cost per task before any long-term commitment.
-
Tech
Facebook lets users steer their own feed
Meta has brought its Your Algorithm feature, already on Instagram, to Facebook. With help from AI, users can say what they want to see more or less of in their feed. The company hopes to answer criticism about opaque recommendations. Control still has limits, since the algorithm remains Meta's.
What it means: Bringing the Your Algorithm feature from Instagram to Facebook answers years of criticism that users do not understand why they see what they see in their feed, and that recommendation systems favour content that holds attention without necessarily being useful. With AI assistance, users can ask for more or less of certain topics, which looks like more control, but in practice it is still a setting inside a system Meta owns and runs. In Europe, the Digital Services Act already requires large platforms to explain their recommendation criteria and offer choices, so features like this are also a way to demonstrate compliance. For businesses that rely on organic reach on Facebook, the change can have practical effects, because if users dial down whole categories of content some posts will reach fewer people, which strengthens the case for owned channels such as a newsletter and a website that the business controls directly.
-
Tech
Cambridge Heavy Industries reveals a desktop fusion device
Cambridge Heavy Industries is developing a small laser-driven nuclear fusion device. The goal is not to generate electricity but to support future fusion power plants. The company worked quietly and is now showing the project publicly. It is one of many new firms betting on the coming fusion industry.
What it means: The Cambridge Heavy Industries device is interesting precisely because it does not promise electricity: instead of chasing net energy from fusion, which remains distant, it aims to be a tool for the people who will build future plants, for example to test materials or sensors in conditions that resemble those inside a reactor. The approach recalls the old strategy of selling shovels in a gold rush, and it is more realistic as a business, since it can bring revenue long before fusion is shown to be commercially viable. The company working quietly and only now going public fits a wider wave of investment in the field, with many new firms building the supply chain for an industry that does not exist yet. For readers outside physics, the takeaway is that headlines about desktop fusion deserve careful reading, because the progress is real but often concerns tools and testing rather than cheap power any time soon.
-
Tech
Bumble accepts a $1.3B buyout as dating apps rethink their model
After years of falling subscribers and revenue, Bumble has accepted a takeover offer worth $1.3 billion. The deal comes as the whole dating app market looks for a new model. Users, especially younger ones, are drifting away from classic swiping. The price shows how far the company's valuation has fallen.
What it means: Accepting a 1.3 billion dollar takeover closes a chapter for a company once seen as one of the most dynamic in dating apps, and shows how far valuations across the sector have fallen after years of shrinking subscribers and revenue. The core problem is not Bumble's alone: endless swiping tires users, especially younger ones, who want more genuine connections, less pressure and more safety, while paid subscriptions look harder to justify when the experience does not improve. The market is now testing alternatives such as events, group activities and tools that get people to a real meeting faster, and the new owners will be judged on whether they invest in that kind of change or simply cut costs. For anyone building new dating products, the deal is both a warning and an opportunity, because the audience is still there, but it does not want one more app that looks like all the others.
-
Tech
Apple opens orders for HomeView, a $299 smart display
Apple has opened preorders for HomeView, a 7-inch smart display with a speaker priced at $299. It acts as a control hub for the smart home and can recognise up to six family members through its camera. Users can ask Siri to close blinds or lock doors by voice. It can also show up to nine security camera feeds at once.
What it means: HomeView is Apple's first smart display and arrives in a category where Amazon and Google have been present for years, which means the company is betting not on a new idea but on integration with the rest of its ecosystem and on its reputation for being more careful with privacy. The features described, voice control for blinds and locks and up to nine security cameras on screen at once, position the device as a home control hub rather than just a speaker with a display. The most sensitive element is recognising up to six family members through a camera, because that is biometric data inside the most private space, and user trust will depend on where that data is processed and who can reach it. In Europe, where GDPR treats biometric information strictly, Apple will need to explain clearly how recognition works, while for smart home device makers the entry of such a large player raises the bar on security and ease of use.
-
Tech
Apple updates the HomePod mini with a new chip at $149
Apple refreshed the HomePod mini with a newer processor and support for Siri AI. The speaker also gains a lighting effect called SoundGlow. The price rises to $149, which is $50 more than before. It goes on sale on October 30 as part of Apple's wider smart home push.
What it means: The new HomePod mini is not just a chip refresh, it is part of Apple's wider effort to restart its smart home presence, with Siri AI as the main reason for owners of the previous model to upgrade. The price rising to 149 dollars, fifty more than before, is telling: Apple no longer treats the small speaker as a cheap way in, but as a device that has to carry the new artificial intelligence features, which needs more capable hardware. The SoundGlow lighting effect is mostly an aesthetic addition, but it shows the company caring about the experience inside the home and not only about sound. For consumers, the question is whether the voice assistant improvements are substantial enough to justify the higher price, while for anyone building smart home devices or apps, Apple's investment means compatibility with its ecosystem will matter more over the coming months, both for new products and for updates to existing ones.
-
Tech
AI labs pay for the code and records of failed startups
AI companies are paying for the code and files of startups that are shutting down, to train their models. One founder closing his company received around $25,000 for the codebase alone. Such data is valued because it is real and not publicly available. The practice raises questions about the privacy of the former customers.
What it means: Buying data from companies that are shutting down shows how much the value of a startup's code and records has changed: even when the product fails, the source code, documentation and internal files are real, non-public material that AI labs use to improve their models on genuine programming problems. Around 25,000 dollars for one company's code is not a fortune, but for a founder winding down it can cover liquidation costs, which creates a strong incentive to sell as many files as possible. That is exactly where the risk lies, because if those files include customer details, support messages or logs, selling them may breach the terms of service or GDPR, since users handed over their data for a different purpose. For any founder considering such a sale, the right first step is a legal review of what is allowed and a clean separation of code from personal data, and for customers of small vendors it is a reminder to check what happens to their data if the provider stops operating.
-
Tech
Drone strikes expose how fragile Russia's data centers are
After Ukrainian drone strikes hit Russian data centers, including Yandex facilities, Russian bloggers are warning about the country's vulnerability. Russia's digital economy is concentrated in too few sites. The outages disrupted services used by millions of people. The episode shows that digital infrastructure is now a military target.
What it means: Drone strikes on Russian data centers, including Yandex facilities, bring forward something that is easy to forget: digital services we think of as weightless sit in specific buildings with specific power and cooling, and when those are hit, millions of people lose access to payment apps, maps or messaging. Warnings from Russian bloggers about infrastructure concentrated in too few places show that the problem is architectural as well as military, since a system without geographic spread has a single point of failure. For businesses in peacetime the threat may be fire, flooding, a power cut or an outage at one provider, but the conclusion is the same: critical data and services need copies in a different location and a recovery plan that has actually been tested. European rules such as DORA for the financial sector already require this kind of planning, and the experience of war gives practical weight to requirements some had dismissed as bureaucracy.
Summaries are ours. Copyright in the original articles remains with their publishers.
Have a project in mind?
Tell us what you need to achieve. You get a concrete proposal with scope, timeline and cost — no obligation.