NASA is planning to host a live Twitch stream next week from the International Space Station (ISS). "The stream, which takes place on February 12th at 11:45AM ET on NASA's Twitch channel, will feature Don Pettit, an astronaut currently on the ISS, and Matt Dominick, who returned to Earth from the ISS in October," reports The Verge. From the report: The astronauts will discuss "daily life aboard the space station and the research conducted in microgravity" and viewers will be able to ask them questions, according to a blog post.

"This Twitch event from space is the first of many," Brittany Brown, director of the Office of Communications Digital and Technology Division at NASA Headquarters in Washington, says in the post. "We spoke with digital creators at TwitchCon about their desire for streams designed with their communities in mind, and we listened. In addition to our spacewalks, launches, and landings, we'll host more Twitch-exclusive streams like this one."

Google has begun internal testing of a new "AI Mode" for its search engine, powered by its Gemini 2.0 AI model, according to a company email seen by technology news site 9to5Google. The feature, which appears alongside existing filters like Images and News, creates a chatbot-like interface for handling complex queries and follow-up questions.

It generates detailed responses with web links displayed in a card format on the right side of the screen. AI Mode targets exploratory searches such as product comparisons and how-to questions that traditional search results may not effectively address. The company is currently testing the feature with U.S.-based employees, with CEO Sundar Pichai indicating a possible launch this year.

The Entertainment Software Association is launching a new gaming event to replace E3, which was permanently canceled in 2023. According to Engadget, the new event is called iicon (short for "interactive innovation conference") and will feature many of the same major gaming companies that once participated in E3. "Sony, Nintendo, Microsoft, Disney, EA, Epic Games, Ubisoft, Square Enix, Take Two Interactive, Amazon Games and Warner Bros. Games are all named as participants." From the report: [T]he announcements on social media promote iicon as being for "visionaries," "changemakers" and "innovators," so our best guess is that this event will swing more toward the corporate side of gaming where people might use that language unironically. If that's the case, this won't really be a replacement for the heyday of E3, when studios big and small would showcase their upcoming projects and drop internet-breaking surprises. Instead, the inaugural event in April 2026 sounds like it will focus more on moving the needle, brand alignments and synergy.

An anonymous reader quotes a report from Ars Technica: On Tuesday, Hugging Face researchers released an open source AI research agent called "Open Deep Research," created by an in-house team as a challenge 24 hours after the launch of OpenAI's Deep Research feature, which can autonomously browse the web and create research reports. The project seeks to match Deep Research's performance while making the technology freely available to developers. "While powerful LLMs are now freely available in open-source, OpenAI didn't disclose much about the agentic framework underlying Deep Research," writes Hugging Face on its announcement page. "So we decided to embark on a 24-hour mission to reproduce their results and open-source the needed framework along the way!"

Similar to both OpenAI's Deep Research and Google's implementation of its own "Deep Research" using Gemini (first introduced in December -- before OpenAI), Hugging Face's solution adds an "agent" framework to an existing AI model to allow it to perform multi-step tasks, such as collecting information and building the report as it goes along that it presents to the user at the end. The open source clone is already racking up comparable benchmark results. After only a day's work, Hugging Face's Open Deep Research has reached 55.15 percent accuracy on the General AI Assistants (GAIA) benchmark, which tests an AI model's ability to gather and synthesize information from multiple sources. OpenAI's Deep Research scored 67.36 percent accuracy on the same benchmark with a single-pass response (OpenAI's score went up to 72.57 percent when 64 responses were combined using a consensus mechanism).

As Hugging Face points out in its post, GAIA includes complex multi-step questions such as this one: "Which of the fruits shown in the 2008 painting 'Embroidery from Uzbekistan' were served as part of the October 1949 breakfast menu for the ocean liner that was later used as a floating prop for the film 'The Last Voyage'? Give the items as a comma-separated list, ordering them in clockwise order based on their arrangement in the painting starting from the 12 o'clock position. Use the plural form of each fruit." To correctly answer that type of question, the AI agent must seek out multiple disparate sources and assemble them into a coherent answer. Many of the questions in GAIA represent no easy task, even for a human, so they test agentic AI's mettle quite well. Open Deep Research "builds on OpenAI's large language models (such as GPT-4o) or simulated reasoning models (such as o1 and o3-mini) through an API," notes Ars. "But it can also be adapted to open-weights AI models. The novel part here is the agentic structure that holds it all together and allows an AI language model to autonomously complete a research task."

The code has been made public on GitHub.

In his post about the future of Fedora Workstation, Christian F.K. Schaller discusses how the Red Hat team plans to integrate AI with IBM's open-source Granite engine to enhance developer tools, such as IDEs, and create an AI-powered Code Assistant. He says the team is also working on streamlining AI acceleration in Toolbx and ensuring Fedora users have access to tools like RamaLama. From the post: One big item on our list for the year is looking at ways Fedora Workstation can make use of artificial intelligence. Thanks to IBMs Granite effort we know have an AI engine that is available under proper open source licensing terms and which can be extended for many different usecases. Also the IBM Granite team has an aggressive plan for releasing updated versions of Granite, incorporating new features of special interest to developers, like making Granite a great engine to power IDEs and similar tools. We been brainstorming various ideas in the team for how we can make use of AI to provide improved or new features to users of GNOME and Fedora Workstation. This includes making sure Fedora Workstation users have access to great tools like RamaLama, that we make sure setting up accelerated AI inside Toolbx is simple, that we offer a good Code Assistant based on Granite and that we come up with other cool integration points. "I'm still not sure how I feel about this approach," writes designer/developer and blogger, Bradley Taunt. "While IBM Granite is an open source model, I still don't enjoy so much artificial 'intelligence' creeping into core OS development. This also isn't something optional on the end-users side, like a desktop feature or package. This sounds like it's going to be built directly into the core system."

"Red Hat has been pushing hard towards AI and my main concern is having this influence other operating system dev teams. Luckily things seems AI-free in BSD land. For now, at least."

Cloudflare, a major web infrastructure company, will now track and verify the authenticity of images across its network through Content Credentials, a digital signature system that documents an image's origin and editing history. The technology, developed by Adobe's Content Authenticity Initiative, embeds metadata showing who created an image, when it was taken, and any subsequent modifications - including those made by AI tools.

Major news organizations including the BBC, Wall Street Journal and New York Times have already adopted the system. The feature is available immediately through a single toggle in Cloudflare Images settings. Users can verify an image's authenticity through Adobe's web tool or Chrome extension.

"iPhone devices are now eligible to test SpaceX-owned Starlink's direct-to-cell capability," Reuters reported this week, citing an announcement from T-Mobile: T-Mobile and Elon Musk's SpaceX are currently testing the Starlink cell network on a trial basis after receiving approval from the Federal Communications Commission in November last year. The trial offers 'text via satellite', while voice and data features will be added in the future, according to the T-Mobile website. T-Mobile initially only listed a few Android smartphones as eligible devices to test the network, but has now added iPhone devices with the latest iOS 18.3 software update.
The next day stock prices fell for several direct-to-smartphone satellite companies, reports SpaceNews: Shares in Globalstar, which enables connectivity beyond the reach of cellular towers on the latest iPhones via a far-reaching partnership with Apple, closed down nearly 18% the following day. Constellation developer AST SpaceMobile slipped 12%. Canada's MDA, which is building at least 17 satellites for Globalstar after Apple agreed to cover most of the costs to replenish the constellation, also saw its shares fall more than 9%...

"Combined, today's price action in Globalstar and satellite manufacturer MDA suggest a real investor fear that SpaceX could disintermediate the Apple-Globalstar partnership," said Adam Rhodes, a senior telecoms analyst at Octus. "However, it appears to us that there is room for both services. Based on the information we have seen, we do not anticipate that Apple views the T-Mobile-Starlink service as a replacement for the Globalstar MSS network, but rather it is choosing to enable the added feature on its T-Mobile phones...." B. Riley analyst Mike Crawford noted that Apple's two binding contracts with Globalstar extend well into the next decade, ensuring both capital expenditure (capex) and recurring service revenues.
Thanks to Slashdot reader jjslash for sharing the news.

In 2023, Microsoft began including a free VPN feature in its "Microsoft Defender" security app for all Microsoft 365 subscribers ("Personal" and "Family"). Originally Microsoft had "called it a privacy protection feature," writes the blog Windows Central, "designed to let you access sensitive data on the web via a VPN tunnel." But.... Unfortunately, Microsoft has now announced that it's killing the feature later this month, only a couple of years after it first debuted...

To add insult to injury, this announcement comes just days after Microsoft increased subscription prices across the board. Both Personal and Family subscriptions went up by three dollars a month, which the company says is the first price hike Microsoft 365 has seen in over a decade. The increased price does now include Microsoft 365 Copilot, which adds AI features to Word, PowerPoint, Excel, and others.

However, it also comes with the removal of the free VPN in Microsoft Defender, which I've found to be much more useful so far.

Google is enhancing Android 16's Linux Terminal app to support graphical Linux applications, so Android Authority decided to put it to the test by running Doom. From the report: The Terminal app first appeared in the Android 15 QPR2 beta as a developer option, and it still remains locked behind developer settings. Since its initial public release, Google pushed a few changes that fixed issues with the installation process and added a settings menu to resize the disk, forward ports, and backup the installation. However, the biggest changes the company has been working on, which include adding hardware acceleration support and a full graphical environment, have not been pushed to any public releases.

Thankfully, since Google is working on this feature in the open, it's possible to simply compile a build of AOSP with these changes added in. This gives us the opportunity to trial upcoming features of the Android Linux Terminal app before a public release. To demonstrate, we fired up the Linux Terminal on a Pixel 9 Pro, tapped a new button on the top right to enter the Display activity, and then ran the 'weston' command to open up a graphical environment. (Weston is a reference implementation of a Wayland compositor, a modern display server protocol.)

We also went ahead and enabled hardware acceleration beforehand as well as installed Chocolate Doom, a source port of Doom, to see if it would run. Doom did run, as you can see below. It ran well, which is no surprise considering Doom can run on literal potatoes. There wasn't any audio because an audio server isn't available yet, but audio support is something that Google is still working on.

Google users have discovered that adding curse words to search queries disables the company's AI-powered overview feature. While Google's Gemini AI system typically avoids profanity, inserting expletives into search terms bypasses AI summaries and delivers traditional web results instead. Users can also disable AI overviews by adding "-ai" or other text strings after a minus sign to their queries.

jjslash shares a report from TechSpot: Atari teamed up with luxury watch brand Nubeo to celebrate the 45th anniversary of Asteroids with a collection of five limited-edition timepieces. Each watch, originally priced at $1,650 but discounted to $499, was limited to 125 pieces -- and they sold out almost immediately. The watches feature a unique Japanese automatic movement, where three rotating discs replace traditional hands. The smallest disc, featuring the classic Asteroids spaceship, acts as the second hand, while the minute and hour hands are represented by asteroid-filled outer discs.

While they're not smartwatches, the timepieces feature Swiss Super-LumiNova glow-in-the-dark ink sitting underneath a sapphire lens within a stainless-steel case. They're water resistant up to 21 ATM (atmospheres) and have a screw-down crown, so you can show them off while at the beach or diving.

A new update to CVS's mobile app includes a feature that allows some customers to access items on locked shelves using their phone -- "without having to summon an overworked employee to open it first," reports The Verge. The feature is currently being trialed in a handful of stores, but will be expanded to many more locations later this year if it goes well. From the report: According to The Wall Street Journal, "app users need to be logged in, on the local store Wi-Fi, and with their device's Bluetooth enabled to activate the feature." You've also got to be a member of the CVS loyalty program if you want the convenience of grabbing secured merchandise without calling for help. Signing up for that gives CVS plenty of insight into your shopping habits, so keep that in mind as you weigh the convenience of not waiting around.

"People really, really dislike locked cabinets," Tilak Mandadi, executive vice president of ventures at CVS Health, told the Journal. Walmart has apparently come to the same realization, as the massive US retailer conducted a similar test last year. CVS aims to expand the program to around 15 stores soon and eventually reach national availability if all goes well.

An anonymous reader shares a report: LinkedIn has removed at least two accounts that were created for AI "co-workers" whose profile images said they were "#OpenToWork." "I don't need coffee breaks, I don't miss deadlines, and I'll outperform any social media team you've ever worked with -- Guaranteed," the profile page for one of these AI accounts called Ella said. "Tired of human 'experts' making excuses? I deliver, period." The #OpenToWork flair on profile pictures is a feature on LinkedIn that lets people clearly signal they are looking for a job on the professional networking platform.

"People expect the people and conversations they find on LinkedIn to be real," a LinkedIn spokesperson told me in an email. "Our policies are very clear that the creation of a fake account is a violation of our terms of service, and we'll remove them when we find them, as we did in this case." The AI profiles were created by an Israeli company called Marketeam, which offers "dedicated AI agents" that integrate with a client's marketing team and help them execute their marketing strategies "from social media and content marketing to SEO, RTM, ad campaigns, and more."

Google says it will end Chrome Sync support for browser versions more than four years old starting in early 2025. Users running outdated Chrome versions will see error messages prompting them to update their browsers to maintain access to synced data across devices. Those unable to update to newer versions will permanently lose the syncing feature, according to the firm.

An anonymous reader quotes a report from Ars Technica: On Thursday, Anthropic announced Citations, a new API feature that helps Claude models avoid confabulations (also called hallucinations) by linking their responses directly to source documents. The feature lets developers add documents to Claude's context window, enabling the model to automatically cite specific passages it uses to generate answers. "When Citations is enabled, the API processes user-provided source documents (PDF documents and plaintext files) by chunking them into sentences," Anthropic says. "These chunked sentences, along with user-provided context, are then passed to the model with the user's query."

The company describes several potential uses for Citations, including summarizing case files with source-linked key points, answering questions across financial documents with traced references, and powering support systems that cite specific product documentation. In its own internal testing, the company says that the feature improved recall accuracy by up to 15 percent compared to custom citation implementations created by users within prompts. While a 15 percent improvement in accurate recall doesn't sound like much, the new feature still attracted interest from AI researchers like Simon Willison because of its fundamental integration of Retrieval Augmented Generation (RAG) techniques. In a detailed post on his blog, Willison explained why citation features are important.

"The core of the Retrieval Augmented Generation (RAG) pattern is to take a user's question, retrieve portions of documents that might be relevant to that question and then answer the question by including those text fragments in the context provided to the LLM," he writes. "This usually works well, but there is still a risk that the model may answer based on other information from its training data (sometimes OK) or hallucinate entirely incorrect details (definitely bad)." Willison notes that while citing sources helps verify accuracy, building a system that does it well "can be quite tricky," but Citations appears to be a step in the right direction by building RAG capability directly into the model. Anthropic's Alex Albert clarifies that Claude has been trained to cite sources for a while now. What's new with Citations is that "we are exposing this ability to devs." He continued: "To use Citations, users can pass a new 'citations [...]' parameter on any document type they send through the API."

Meta's AI chatbot will now use personal data from users' Facebook and Instagram accounts for personalized responses in the United States and Canada, the company said in a blog post. The upgraded Meta AI can remember user preferences from previous conversations across Facebook, Messenger, and WhatsApp, such as dietary choices and interests. CEO Mark Zuckerberg said the feature helps create personalized content like bedtime stories based on his children's interests. Users cannot opt out of the data-sharing feature, a Meta spokesperson told TechCrunch.

The EFF has complained that in general "smart" products for babies "collect a ton of information about you and your baby on an ongoing basis". (For this year's "worst in privacy" product at CES they chose a $1,200 baby bassinet equipped with a camera, a microphone, and a radar sensor...)

But today the Washington Post reported on a $1,700 bassinet that surprised the mother of a one-month-old when it "abruptly demanded money for a feature she relied on to soothe her baby to sleep." The internet-connected bassinet... reliably comforted her 1-month-old — just as it had her first child — until it started charging $20 a month for some abilities, including one that keeps the bassinet's motion and sounds at one level all night. The level-lock feature previously was available without a fee. "It all felt really intrusive — like they went into our bedroom and clawed back this feature that we've been depending on...." When the Snoo's maker, Happiest Baby, introduced a premium subscription for some of the bassinet's most popular features in July, owners filed dozens of complaints to the Federal Trade Commission and the Better Business Bureau, coordinated review bombs and vented on social media — saying the company took advantage of their desperation for sleep to bait-and-switch them...

Happiest Baby isn't the only baby gear company that has rolled out a subscription. In 2023, makers of the Miku baby monitor, which retails for up to $400, elicited similar fury from parents when it introduced a $10 monthly subscription for most features. A growing number of internet-connected products have lost software support or functionality after purchase in recent years, such as Spotify's Car Thing — a $90 Bluetooth streaming device that the company announced in May it plans to discontinue — and Levi's $350 smart jacket, which let users control their phones by swiping sensors on its sleeve...

Seventeen consumer protection and tech advocacy groups cited Happiest Baby and Car Thing in a letter urging the FTC to create guidelines that ensure products retain core functionality without the imposition of fees that did not exist when the items were originally bought.
The Times notes that the bassinets are often resold, so the subscription fees are partly to cover the costs of supporting new owners, according to Happiest Baby's vice president for marketing and communications. But the article three additional perspectives:

• "This new technology is actually allowing manufacturers to change the way the status quo has been for decades, which is that once you buy something, you own it and you can do whatever you want. Right now, consumers have no trust that what they're buying is actually going to keep working." — Lucas Gutterman, who leads the Public Interest Research Group's "Design to Last" campaign.

• "It's a shame to be beholden to companies' goodwill, to require that they make good decisions about which settings to put behind a paywall. That doesn't feel good, and you can't always trust that, and there's no guarantee that next week Happiest Baby isn't going to announce that all of the features are behind a paywall." — Elizabeth Chamberlain, sustainability director at iFixit.

• "It's no longer just an out-and-out purchase of something. It's a continuous rental, and people don't know that." — Natasha Tusikov, an associate professor at York University

This weekend Cory Doctorow delved into "the two factors that make services terrible: captive users, and no constraints." If your users can't leave, and if you face no consequences for making them miserable (not solely their departure to a competitor, but also fines, criminal charges, worker revolts, and guerrilla warfare with interoperators), then you have the means, motive and opportunity to turn your service into a giant pile of shit... Every economy is forever a-crawl with parasites and monsters like these, but they don't get to burrow into the system and colonize it until policymakers create rips they can pass through.
Doctorow argues that "more and more critics are coming to understand that lock-in is the root of the problem, and that anti-lock-in measures like interoperability can address it." Even more important than market discipline is government discipline, in the form of regulation. If Zuckerberg feared fines for privacy violations, or moderation failures, or illegal anticompetitive mergers, or fraudulent advertising systems that rip off publishers and advertisers, or other forms of fraud (like the "pivot to video"), he would treat his users better. But Facebook's rise to power took place during the second half of the neoliberal era, when the last shreds of regulatory muscle that survived the Reagan revolution were being devoured... But it's worse than that, because Zuckerberg and other tech monopolists figured out how to harness "IP" law to get the government to shut down third-party technology that might help users resist enshittification... [Doctorow says this is "why companies are so desperate to get you to use their apps rather than the open web"] IP law is why you can't make an alternative client that blocks algorithmic recommendations. IP law is why you can't leave Facebook for a new service and run a scraper that imports your waiting Facebook messages into a different inbox. IP law is why you can't scrape Facebook to catalog the paid political disinformation the company allows on the platform...
But then Doctorow argues that "Legacy social media is at a turning point," citing as "a credible threat" new systems built on open standards like Mastodon (built on Activitypub) and Bluesky (built on Atproto): I believe strongly in improving the Fediverse, and I believe in adding the long-overdue federation to Bluesky. That's because my goal isn't the success of the Fediverse — it's the defeat of enshtitification. My answer to "why spend money fixing Bluesky?" is "why leave 20 million people at risk of enshittification when we could not only make them safe, but also create the toolchain to allow many, many organizations to operate a whole federation of Bluesky servers?" If you care about a better internet — and not just the Fediverse — then you should share this goal, too... Mastodon has one feature that Bluesky sorely lacks — the federation that imposes antienshittificatory discipline on companies and offers an enshittification fire-exit for users if the discipline fails. It's long past time that someone copied that feature over to Bluesky.
Doctorow argues that federated and "federatable" social media "disciplines enshittifiers" by freeing social media's captive audiences.

"Any user can go to any server at any time and stay in touch with everyone else."

ZDNET's senior editor sees an "overwhelmingly negative" response to Microsoft's surprise price hike for the 84 million paying subscribers to its Microsoft 365 software suite. Attempting the first price hike in more than 12 years, "they made it a 30% price increase" — going from $10 a month to $13 a month — "and blamed it all on artificial intelligence." Bad idea. Why? Because...

No one wants to pay for AI...

If you ask Copilot in Word to write something for you, the results will be about what you'd expect from an enthusiastic summer intern. You might fare better if you ask Copilot to turn a folder full of photos into a PowerPoint presentation. But is that task really such a challenge...?

The announcement was bungled, too... I learned about the new price thanks to a pop-up message on my Android phone... It could be worse, I suppose. Just ask the French and Spanish subscribers who got a similar pop-up message telling them their price had gone from €10 a month to €13,000. (Those pesky decimals.) Oh, and I've lost count of the number of people who were baffled and angry that Microsoft had forcibly installed the Copilot app on their devices. It was just a rebranding of the old Microsoft 365 app with the new name and logo, but in my case it was days later before I received yet another pop-up message telling me about the change...

[T]hey turned the feature on for everyone and gave Word users a well-hidden checkbox that reads Enable Copilot. The feature is on by default, so you have to clear the checkbox to make it go away. As for the other Office apps? "Uh, we'll get around to giving you a button to turn it off next month. Maybe." Seriously, the support page that explains where you can find that box in Word says, "We're working on adding the Enable Copilot checkbox to Excel, OneNote, and PowerPoint on Windows devices and to Excel and PowerPoint on Mac devices. That is tentatively scheduled to happen in February 2025." Until the Enable Copilot button is available, you can't disable Copilot.
ZDNET's senior editor concludes it's a naked grab for cash, adding "I could plug the numbers into Excel and tell you about it, but let's have Copilot explain instead."

Prompt: If I have 84 million subscribers who pay me $10 a month, and I increase their monthly fee by $3 a month each, how much extra revenue will I make each year?

Copilot describes the calculation, concluding with "You would make an additional $3.024 billion per year from this fee increase." Copilot then posts two emojis — a bag of money, and a stock chart with the line going up.

As spotted by Tom's Hardware, Nvidia's CUDA 12.8 release notes signal the transition of Maxwell, Pascal, and Volta GPUs to the legacy driver branch. As a result, there will be no more new feature updates for these architectures; however, CUDA and gaming driver support will remain for now. From the report: It's crucial to highlight that this has nothing to do with GeForce gaming driver support. In fact, Maxwell and Pascal continue to be on the support list for the GeForce RTX series driver, unlike Kepler. Nvidia didn't detail whether or when it'll drop support for Maxwell, Pascal, and Volta GPUs for the gaming driver.

Nvidia has not issued an exact date for the end of full support for these three GPU architectures, but it will soon. The current CUDA toolkit still supports the three affected architectures, but they won't receive future updates. Once the move goes through, the only remaining GTX-series GPUs with full support will be the GTX 16-series, based on the RTX 20-series' Turing architecture.