找到 173 篇相关文章
Cloudflare recently introduced Precursor, a client-side behavioral analysis engine that continuously evaluates session interactions, such as mouse movements and keyboard timing, to improve detection of sophisticated bots and AI agents without relying solely on one-time challenges like CAPTCHAs. By Renato Losio
WIRED spoke with Tucker Bryant, an artist and former Google employee who created ChatTJB to get people to reflect on the “strange moment” we’re in.
If you're building — or integrating — a browser-based audio trimmer, the question that eventually reaches your inbox isn't "does it cut audio?" The real question is: does it cut audio correctly across the inputs we actually receive from users? That shift, from feature presence to behavior under fuzzy conditions, is what turns a demo into a product. This article walks through the QA matrix I use when reviewing client-side trimmers before release, with an emphasis on the silent failures that don't show up in a happy-path recording. The tool under review for most of this article is the Lizely audio cutter ( in-depth walkthrough ), but the principles apply to any browser trimmer that decodes via AudioContext or OfflineAudioContext . What "Trim" Actually Means Once You Leave the Lab In the lab, you upload a 44.1 kHz stereo WAV, drag two handles, click export, and verify the output. In production, users upload M4A recordings from iPhone Voice Memos, AMR files from old Android handsets, mono 8 kHz captures from cheap conference mics, and — occasionally — files renamed from .wav to .mp3 without re-encoding. Each of those paths stresses a different layer of the pipeline. The first thing to test, before any UI work, is the decode step. Browsers expose this through the decodeAudioData method on BaseAudioContext , documented on MDN's BaseAudioContext page . MDN is explicit about something engineers often miss: decodeAudioData detaches the input ArrayBuffer . If your trimmer holds a reference to the original buffer for "undo" and reuses it, you'll decode an empty buffer the second time around and get a silent result. That's a real defect class, not a theoretical one. The second thing to test is what happens when decoding fails. The spec says decodeAudioData invokes the error callback with a DOMException , but the browser-specific error messages vary. Chrome tends to surface "Decoding error" with no detail; Firefox appends the underlying codec name. Your QA suite should assert on
This may help fungi survive host scarcity and explains why infected hosts prefer the mosses as death sites.
A 0.3-second disagreement between two sources of truth made my first paying customer fail four times. The browser preview stored the project duration rounded to a whole second: 3983s. The worker that processed the audio measured the real media: 3982.699–3982.788s. Cue generation ran against the rounded number. Delivery certification ran against the trusted measurement. Any candidate built on the rounded boundary exceeded the certified boundary by 212–301ms — so the final cue failed, deterministically, every single time. That customer ended up with four projects and three distinct audio files — four identical failures, each one blocked by the same gate. No subtitle asset, no explanation, no path forward. No alert fired. No complaint had come in. I found it because I was looking. Here is the part worth writing down: the quality gate did exactly what it was designed to do. It rejected every unsafe result before it could reach the customer. And the customer still lost. Four failures, and not one of them was a gate that misbehaved — they were four places where quality had been treated as a check instead of a product decision. A fail-closed gate is an engineering floor, not a product. Quality is not the final check that rejects bad output; it is the input boundary you commit to, the authority you give each fact, the failure states you design for, and the meaning you attach to your own scores. What follows is the postmortem as an engineering story: four deterministic failures, each one a missing product decision, and the contract I now think every pipeline like this should carry. One fact, two authorities The whole incident starts with a single number. The project duration existed twice: The browser preview rounded it to 3983s . The funded worker measured the actual media as 3982.699–3982.788s . Cue generation used the rounded value. Delivery certification used the trusted measurement. The result: the last cue always ended 212–301ms past the certified boundary, and the fin
This marks the official launch of Zoox's commercial operations.
Waymo has dropped the waitlist for its robotaxi service in Dallas, the latest step in the company's bid to scale its self-driving technology across the United States, U.K., and Europe.
Ati Robotics assembles its robots in India and uses just a few Chinese parts—a strategy that could pay off as the Trump administration cracks down on Chinese humanoids.
And it’s personal information (alternate link ): The exposed data includes an AI-powered therapy app that someone appears to have vibe-coded, notes on meetings, and a dashboard someone made apparently to analyze medical billing data. Exposed chats reportedly include private cryptocurrency wallet keys and personal information like peoples’ addresses. What seems to be the issue is a user setting about data sharing. Anthropic’s position is that it’s not their problem : “We give people control over sharing their Claude conversations publicly, and in keeping with our privacy principles, we do not share chat directories or sitemaps with search engines like Google,” the company said in a statement. “These shareable links are not guessable or discoverable unless people choose to share them themselves. When someone shares a conversation, they are making that content publicly accessible, and like other public web content, it may be archived by third-party services.”...
This is a submission for DEV's Summer Bug Smash: Smash Stories powered by Sentry. ...
No other gadget knows as much about your home as a robot vacuum. They map your space, learn your routines, and many now carry a camera and microphone into every room in your house. As AI gives them a better understanding of what they see, the privacy and security risks these autonomous robots pose are […]
July 2026 is over. The month that opened with AUTONOMOUS 2026 and WAIC 2026 running simultaneously on opposite sides of the Pacific closed with the sector tallying what it built. The number that defines the period is $55.8 billion in robotics funding across H1 - nearly double the prior full-year record. But the more durable signal from this week is operational rather than financial: Neura Robotics has a confirmed deployment date at a Schaeffler facility in December, NVIDIA's simulation-to-real pipeline is now functional at production scale, and five simultaneous shifts are reshaping factory floors right now, not in 2027. The questions that drove the first half of 2026 - does Physical AI work, is the funding real, will the robots actually arrive - are no longer interesting. H2 starts with harder ones. Stats: Value Description $55.8B Robotics funding raised in H1 2026, nearly double the prior annual record $8.6B Humanoid startup funding in H1 2026 alone, 1.8x all of 2025 December 2026 Confirmed first deployment of Neura Robotics humanoids at Schaeffler's German facilities 5 Simultaneous operational shifts reshaping factory floors identified in the mid-2026 analysis Neura Robotics Has a Deployment Date: December 2026 in a Schaeffler Factory Most Physical AI deployment announcements are directional. "We are partnering with X to explore robotics in our facilities" is a press release. A confirmed month and a specific facility is a contract. Neura Robotics confirmed that Schaeffler - one of the key investors in its $1.4 billion Series C alongside Amazon, Nvidia, Qualcomm, and the European Investment Bank - plans to deploy Neura's humanoids in its German facilities in December 2026 . Schaeffler manufactures precision bearings and components for electric vehicles, operating in environments where dimensional tolerances are measured in micrometers. Deploying a humanoid robot in that context is a fundamentally different challenge than warehouse pick-and-place or automotive sequ
A new restriction on foreign-made mobile robots affects robotic vacuums, pool cleaners, and lawn mowers. Here’s what it means for the bots already in your home—and the ones you may never get to buy.
Researchers fear AI is moving too fast, while Mark Zuckerberg is worried about who owns it. Plus: Inside Black Forest Labs’ push into robotics.
Gemini Robotics 2 includes three models, but only one is publicly available right now.
Google DeepMind says the latest version of its Gemini Robotics AI model can "control entire humanoid robots." While the previous model focused on controlling a humanoid robot's upper body, Gemini Robotics 2 now supports "whole-body motions" ranging from its feet to fingertips, according to an announcement on Thursday. The new model will allow humanoid robots […]
What Is Retrieval Augmented Generation (RAG), and Why Does It Make AI So Much Less Confidently Wrong? You know that game show contestant who buzzes in before the host finishes reading the question, shouts "MOUNT EVEREST!" with absolute certainty, and then looks genuinely confused when the correct answer turns out to be "the Treaty of Westphalia"? That's been AI for most of its existence. Supremely confident, occasionally correct, and deeply committed to whatever pops into its head first. Now imagine that same contestant gets a new rule: before answering, they can phone a friend who has the exact relevant textbook already open to the right page. The friend reads them the actual answer, word for word, and then the contestant puts it in their own words for the judges. Suddenly, our buzzer-happy friend is getting questions right. That phone call is Retrieval Augmented Generation, and it's the reason AI chatbots have gotten weirdly more useful in the past year. The Old Way: Confidently Wrong at 200 Miles Per Hour Traditional large language models (big AI systems trained on tons of text) get trained on enormous dumps of text scraped from the internet, books, and whatever else researchers can feed them. Then the training ends. The model gets sealed off, frozen in time with whatever it learned. When you ask a question, these models generate answers by predicting the most plausible-sounding next words based on patterns they memorized during training. It's essentially very sophisticated autocomplete. The AI has no fact-checking mechanism. It doesn't "know" anything in the way you know your own phone number. It just knows what words tend to follow other words. This leads to what researchers politely call hallucinations, which is a fancy term for "making stuff up with tremendous confidence." The AI generates text that sounds authoritative and well-structured because it's learned the pattern of how authoritative text sounds. But the actual facts? Those might be completely invent
The latest version of Google DeepMind's AI model includes a significant jump into “physical AGI.” But plopping AI into the real world comes with risks.
Federal safety regulators have given Zoox a temporary exemption that will allow the Amazon-owned autonomous vehicle technology company to charge customers for rides in its custom-built robotaxi.
A German startup sent a camera-wearing chef to my apartment. In exchange for a free lunch, I let them record every chop and stir to train future humanoids.