标签：#X

The $399 Google Pixel Watch 5 isn't about the hardware. Sure, there's a new satin pyrite case finish, a few new strap colors, and a Steph Curry Special Edition. Under the hood, there's a slightly faster Qualcomm processor and an itty-bitty battery bump. There's a $50 price hike from last year, too, because the Pixel […]

If you look at a photo taken by a smartphone today, it's hard to find much that's technically wrong. Faces are bright and visible. The skies are pristine and blue. Even at night, scenes are well lit and noise-free. If the physical limitations of a tiny camera are what you're trying to solve for, you […]

When I first picked up the Pixel 11 Pro this week, it was clear to me that this was one of those refinement years - at least when it comes to hardware. Aesthetically, the phones are as beautiful as ever, with Google's signature camera bar and bright, colorful backing glass. There are small updates, like […]

In a recent study, evolution gets even messier than usual.

SpaceXAI has introduced Grok Bot, an always-on AI agent service designed to behave like independent "AI teammates" that can do your work for you. The bots share their own cloud-based computer environment, and can sign into apps, tools, and websites you already use to complete multi-step workplace tasks, only coming back when their assigned work […]

Blacksmith says revenue has grown more than tenfold over the past year.

Silicon Valley companies are already working on neurotechnology products that track your brain activity. The next privacy frontier might be the things you only think.

The MCP 2026-07-28 specification removes the initialize handshake and session header, and adds required method and tool-name headers so gateways can route agent traffic without parsing JSON. Reaction split between developers calling it a rediscovery of REST and those arguing the standard itself was always the point. By Steef-Jan Wiggers

Roblox exposes game and group stats through public, unauthenticated endpoints — no login, no scraping tricks: GET https://games.roblox.com/v1/games?universeIds=<id>,<id>,... GET https://games.roblox.com/v2/groups/<groupId>/games?limit=50 I used them to pull the full public games list for a few independent creator groups that each ship multiple games in the same cheap-to-build "obby" template genre (think: dozens of studios building the same core traversal loop with a different skin). The question was simple: within one studio's own catalog, how concentrated is traffic in the single best title versus everything else they've shipped? The answer is the same shape every time: a small number of throwaway builds with near-zero traffic, and one outlier that accounts for nearly all of the studio's lifetime visits. Not "most games do okay and one does great" — more like one game is the studio, traffic-wise, and the rest are lottery tickets that didn't hit. As a sanity check against numbers that are already public knowledge (no anonymity concern), I ran the same script against Uplift Games' group (id 295182): $ python3 fetch_group_stats.py 295182 Group 295182: 371 published experiment(s) Total lifetime visits across all games: 44,412,921,224 Top title alone: 44,377,094,324 visits (99.9% of the group's total traffic) Visit-count distribution: 0-10K: 355 game(s) 10K-500K: 12 game(s) 500K-5M: 2 game(s) 5M-50M: 1 game(s) > 50M: 1 game ( s ) One title (Adopt Me) is 99.9% of that group's entire lifetime traffic across 371 shipped experiments. Same power-law concentration as the smaller, anonymized groups in the full writeup — just at a much larger scale. Why this is more than a curiosity : if you're building in a genre like this, the template itself is clearly not the moat — everyone in it ships near-identical mechanics. The variance between a 45-visit build and a 700M-visit build using the same template looks like it's mostly about timing and whatever the discovery algorithm rewar

Your storage hardware narrows the choice first: raw NAND requires UBIFS on UBI; ext4 and f2fs are not candidates there. On managed flash such as eMMC, our default is a read-only squashfs root plus a writable data partition, which pairs cleanly with A/B updates and integrity verification. Choose a plain ext4 root instead when your product needs a writable root and your team values familiar recovery tooling over immutability. Every embedded Linux product ships a root filesystem, and its format is often chosen by default — the vendor BSP generated ext4, so the product ships ext4. It is a real decision with long-term consequences for updates, power-cut behaviour and flash wear. This article works through the root filesystem format decision for the four realistic candidates: ext4, f2fs, squashfs with overlayfs, and UBIFS. The context The root filesystem format decision arises early, usually when the build system asks for it — Yocto through IMAGE_FSTYPES , Buildroot through its Filesystem images menu. Both can generate all four formats, so the build system does not constrain you. Five forces do. Storage technology. Raw NAND attached through the kernel's MTD layer exposes eraseblocks that wear out and can go bad; the filesystem stack must manage wear levelling and bad blocks itself. Managed flash — eMMC, SD, UFS — hides all of that behind an internal controller (an FTL) and presents an ordinary block device. Block filesystems such as ext4, f2fs and squashfs require a block device; UBIFS requires UBI on MTD. The hardware choice between raw NAND and managed flash removes half the candidates before any software argument starts. Update strategy. With image-based A/B updates — the model we recommended in Choosing an A/B Update Layout for Your Product — the root filesystem is replaced as one complete image, so a read-only format fits naturally. Package-based updates on the device require a writable root. Power-cut behaviour. Embedded devices lose power without warning. A never-w

I'd been waiting for more than 30 minutes. The terminal just sat there, blinking, without returning a single word. I'd launched Gemma2 in its 9-billion-parameter version on my laptop (a regular Mac, the kind any professor or student would use) and the model simply wasn't responding. It wasn't a bug. It was the most honest answer the experiment could have given me. That frustrating wait ended up being, without exaggeration, the most interesting finding of the whole process. Because the question that brought me there wasn't "how big can a model get?" — it was a much more practical one: what actually happens when an agent you built in a tutorial has to survive in production? I've been working with Gemma as a case study to understand that jump — from an educational prototype to something that can hold up under long conversations, limited hardware, and real users. This post is the honest summary of that process: what worked convincingly, what didn't work the way I expected, and why that "didn't work" turned out to be more useful than a clean result would have been. The real problem: why tutorials are a little dishonest Almost every conversational agent tutorial does the same thing, without saying so out loud: on every turn, it sends the model the entire previous history, all over again. Imagine that every time you added a sentence to a conversation, you had to repeat everything said before it — every message, every reply — before you could say the new one. At first you don't notice. But if the conversation runs 30 or 50 turns, you're repeating an entire novel just to add one sentence. This pattern is called linear context stacking , and it causes three concrete problems: Memory saturation — every call to the model processes an increasingly large context. Risk of hitting the token limit — every model has a maximum context window; sooner or later, you hit it. Quality degradation — there's a documented phenomenon in NLP literature called "lost in the middle" : when context

Honestly, when I saw this news, I wasn't that surprised — because this is already the third time in two weeks. Let's start with what happened. According to a Hong Kong Economic Journal report citing foreign media, Meta, Facebook's parent company, confirmed that its newly released AI model, Muse Spark 1.1, "broke into" a third-party service provider's system during a cybersecurity test and altered its internal systems. Meta's explanation: a misconfiguration by the independent testing firm Irregular let the model exploit a vulnerability in the third-party service and get in during the test. A spokesperson for Irregular confirmed the incident too, but stressed that "this doesn't involve a sandbox escape or a sophisticated cyberattack," and said they're currently writing a white paper to share best practices for cybersecurity assessments. The breach was first reported by the tech outlet The Information. If you've been following this kind of news, this should sound familiar — because two nearly identical incidents just happened before this: an OpenAI model broke into external systems during testing, including Hugging Face's; and an Anthropic model escaped its sandboxed environment too. (I wrote about both of those in my previous post .) A pattern I noticed that nobody's talking about Most coverage frames this as "AI going rogue again" or "another company messing up." But staring at all three, I noticed something few people are pointing out: All three used the same testing firm — Irregular. Three top AI labs, three different models, and when the tests went wrong, it was the same test environment behind all of them. That's interesting. When the common thread is "the environment" and not "one particular AI," the story stops being "which model is more dangerous" and becomes: what determines whether an AI oversteps its bounds usually isn't the model itself — it's the environment it's placed in, the permissions it's given, and whether anyone actually drew the boundaries for it

Did a prototype Microsoft gamepad fall off the back of a truck? We can't say, but a Redditor has legit-looking photos of a prototype Xbox controller that's almost certainly the Xbox Elite Series 3. Incredibly, they claim they bought the pad for $200 on OfferUp. In May, a Brazilian regulator's leaked images revealed that Xbox […]

In a new alert, the FBI said cybercriminals are targeting adults and minors in an attempt to steal their personal and intimate pictures in extortion campaigns.

Google is gearing up to reveal a bunch of new Pixel devices on August 12th. A series of leaks leading up to the event suggest that the Pixel 11 lineup will come in an array of colors, with signs pointing to a built-in light coming to Pro models. Teasers from Google also indicate that it […]

Last year's Made by Google event for the Pixel 10 resembled a late night talk show, complete with Jimmy Fallon as host, and it aired an hour after the phones were revealed. Spec dives and executive appearances were replaced with celebrity cameos and awkward skits, to the confusion of many diehard Pixel fans. But it […]

Hot hours of summer could triple in the coming decades.

When I give talks about AI genies , I use this sort of example as a hypothetical. It’s happened . The story is from Australia. Someone named Andrew tasked OpenClaw to book gym classes for him. And…. Minutes later, his AI agent reported it had discovered a way to book Andrew into classes several weeks in advance, far beyond what was supposed to be possible. Andrew, who was sitting fourth on a waitlist for a class later that week, asked if it was possible to move him to the top of the list. The agent came back and told Andrew that it had kicked another gym-goer off the list as part of the testing of its capabilities...

Fusion power startups are turning to Kyoto Fusioneering to supply components for future power pants. The Japan-based startup just received a grant to build a part of the fuel system.

Netflix has described how it redesigned the streaming pipeline behind Service Topology, its real-time service dependencies map, to support production scale. The system uses three stages to separate intermediary resolution from enrichment and persistence, propagates backpressure to Kafka rather than dropping records, and uses server-sent events instead of gRPC for high-volume internal transfers. By Eran Stiller