AI 资讯

OpenAI is overhauling ChatGPT's voice mode with a new model that it says is more like "talking to another person." The new GPT-Live-1 is designed to interrupt you less and will also wait for you to continue speaking if you pause mid-conversation. During a press briefing, OpenAI research lead Kundan Kumar called GPT-Live-1 the company's […]

OpenAI says its new voice mode can speak and listen at the same time, a key ability for live translation.

It allows you to reimagine videos stored in Google Photos using Gemini Omni.

OpenAI's new voice mode for ChatGPT will acknowledge you as you speak, and slow down if you ask it to.

One project space for humans and AI agents Discussion | Link

Android Bench is evolving, and developers can help guide that process.

Amazon is working on a project codenamed 'Moonraker' that would enable the AI assistant to handle multi-step tasks more effectively.

For Paradigm, the technical frontier will stretch beyond its cryptocurrency investment roots. This fund is expected to expand its investment focus to include robotics and AI.

Founded in 2024, Prime Intellect’s goal is to give organizations capabilities to train their own agentic systems without relying on frontier AI labs.

Prompt injection and token savings - #1 in benchmarks Discussion | Link

Go from an empty repository to a live custom domain with HTTPS in about 14 minutes, without manually editing a single DNS record. The post How GitHub Copilot enables zero DNS configuration for GitHub Pages appeared first on The GitHub Blog .

Most Dockerfiles work. That's the problem — "it builds and runs" hides a lot of quiet costs in security, speed, and size that don't announce themselves until an audit, an incident, or a cloud bill does it for them. Here are seven mistakes I see constantly, and what to do instead. 1. Running as root By default, the process in your container runs as root — and if someone breaks out, they're root on a surface they shouldn't be. Add a non-root user and switch to it: RUN useradd --system --uid 10001 appuser USER appuser Cheap, and it closes off a whole category of "well, at least it wasn't root" incidents. 2. FROM some-image:latest latest is not a version — it's "whatever was newest when this happened to build." Two builds a week apart can produce different images with no diff to explain it, and a surprise base upgrade is a fun way to spend a Friday. Pin a specific tag, ideally by digest: FROM node:20.11.1-slim 3. Baking secrets into layers COPY .env . or ARG API_KEY followed by using it — and now the secret lives in an image layer forever , recoverable by anyone who pulls the image, even if a later layer deletes the file. Layers are immutable and additive; you can't delete your way out of a leak. Use build secrets ( --mount=type=secret ) or inject at runtime, never at build. 4. No .dockerignore Without one, COPY . . sweeps your .git directory, local env files, node_modules , and test data into the build context — bloating the image and, worse, potentially baking credentials and history into a layer. A five-line .dockerignore is one of the highest-leverage files in the repo. 5. Layer order that destroys your cache COPY . . RUN npm ci # ← reinstalls on EVERY code change Docker invalidates every layer after the first change. Copy the lockfile and install dependencies before copying the rest of your source, so a one-line code change doesn't trigger a full reinstall. This is a build-speed bug hiding as a style choice. 6. Leaving package manager cruft in the image RUN apt-get

This post was originally published on MatrixTrak.com — the production reliability toolkit for trading bot operators and .NET engineers. When requests hang forever and recycling releases stuck work: why missing timeouts create backlog, how to add budgets safely, and the rollout plan that prevents new incidents.. Most production incidents do not start as "down." They start as waiting. At 09:12 a dependency slows down. Your ASP.NET instances look healthy. CPU is fine. Memory is fine. But requests stop finishing. In-flight count climbs. Connection pools stop turning over. You scale out and it does not help because the new instances just join the waiting. The cost is not subtle. Backlog grows, SLAs fail, and on-call starts recycling processes because it is the only thing that releases the stuck work. Then the incident repeats next week because nothing changed about the waiting. This post gives you a production playbook for .NET: how to set time budgets, wire cancellation, and roll it out without triggering a new outage. Rescuing an ASP.NET service in production? Start at the .NET Production Rescue hub . If you only do three things Write down a total budget per request/job (then enforce it). Set per-attempt timeouts for each dependency and log elapsedMs , timeoutMs , and the decision (retry/stop/fallback). Propagate cancellation end-to-end so work stops (no zombie work after timeouts). Why requests hang forever: infinite waits capture capacity Missing timeouts are not a performance problem. They are a capacity problem. When a call can wait forever, it will eventually wait longer than your system can afford. While it waits, it holds something your service needs to operate: a worker slot, a thread, a connection, a lock, or a request budget. Once enough requests or jobs are holding those resources, the system stops behaving like a service and starts behaving like a queue you did not design. From the outside it looks like "everything is slow." Underneath, you are accumulating

I tried to build an arbitrage bot between Kalshi and Polymarket. Sports seemed to be the easiest because the matcher is relatively easy compared to the other markets (economy, bitcoin, weather, etc.) The matcher worked, we got about ~98% of the sports and e-sports market. But there's barely any sizable arbitrage between Kalshi and Polymarket, and what shows up closes in under 10 seconds. For the Argentina vs Egypt, the match with the disputed VAR call and Argentina's stoppage-time comeback from two goals down. Every price swing on that match, including the two around the VAR call, closed inside about 45 seconds. Total arbitrage opportunity net of fees across the whole match: $439, against $20.8 million moving through Polymarket's market alone and $13.8 million in Kalshi open interest. ( https://dino.markets/blog/argentina-egypt-var-price-gap ) That's not an arbitrage opportunity. That's what an efficient pair of order books looks like once you finally have the tooling to watch them at the same time. I logged this properly over a full day too: 870 cross-venue price gaps in one 24-hour window, median time open about 9 seconds, 96 percent closed inside 30. ( https://dino.markets/blog/how-long-a-mispricing-lasts ). So I shipped the Polymarket-Kalshi sports market matcher as an API instead of an arbitrage trading bot. It turned out the matcher itself was the useful part. Free REST access to the matched feed and the confirmed-arb view, 60 requests a minute, MCP server included so an agent can read it without you writing a client. Planning to open source the matching engine itself at some point. After that, either extend it to other market pairs between Kalshi and Polymarket, or look at arbitrage against traditional sportsbooks. Nothing locked in yet. Feel free to use it and tell me what you think about it. Thanks!

I Built a Platform Where builders Can Document Their Products Before They Even Launch One thing I've learned after building side projects is that writing code isn't the hardest part. Getting people to notice what you've built is. Every time I finished a project, I'd launch it on a few platforms, share it on X, and hope someone would find it. Sometimes I'd get a few users, but after a day or two, the momentum was gone. It made me realize something. Most platforms are designed for the launch, not the journey. But as developers, the journey is where the interesting stuff happens. You fix bugs, redesign the UI three times, celebrate your first user, rewrite your backend, and slowly turn an idea into a real product. Those moments are worth sharing too. So I started building LaunchDock.space . The idea is simple. Instead of only launching finished products, developers can also create a page for projects that are still in development and post daily progress updates. Think of it as a place to build in public, document your progress, and grow an audience before your product is even ready. Along with development logs, LaunchDock lets makers: Launch finished products. Discover tools built by other builders. Follow the progress of other makers. Connect with a community that enjoys discovering new projects. I'm building LaunchDock with React, TypeScript, Node.js, Express, MongoDB, and Cloudflare R2, keeping the stack simple and focused on performance. The project is still evolving, and I'm shipping new features regularly. Building it has taught me a lot about product design, user feedback, and the importance of consistent shipping. I'd love to hear your thoughts. If you were using a platform like this, what feature would make you come back every day?

There are a lot of fast-growing AI startups, but some are growing even faster, they say.

The robotaxi stopped, called 911, and waited for the San Mateo Police to show up.

What if the AI you chat with every day is quietly running something that looks a lot like a train of thought, and we just never had the right tool to see it? On 7th July, 2026, Anthropic published a research paper that honestly feels a little spooky. The team behind the Transformer Circuits Thread released a long, detailed study called Verbalizable Representations Form a Global Workspace in Language Models . The title is dense, but the idea inside is wild. They found a small, privileged region inside Claude and similar models. A region that behaves a lot like what cognitive scientists call the global workspace , the part of the brain associated with conscious access. The part that lets you say, I am thinking about a banana right now . In this post, I want to walk you through what they found, in plain English, with no math fear and no jargon walls. We will cover what the workspace is, how they found it, what they can do with it, and why it matters for anyone building or using AI. Grab a coffee. This one is worth your time. First, a Quick Brain Detour Before we get to the model, we need a tiny bit of background from neuroscience. For decades, scientists have noticed that the brain seems to operate on two tracks. Most of what your brain does, like parsing the sounds coming into your ears or keeping you balanced, happens automatically and quietly. You cannot really talk about it . It just runs in the background. But a smaller slice of brain activity is different. It is reportable . You can put it into words. You can hold a concept in mind, dismiss it, chain it to another concept, and use it for reasoning. Cognitive scientists call this access consciousness . One popular theory, called the Global Workspace Theory , says this happens because the brain has a shared hub. Specialized processors do their own thing in parallel. But every now and then, a representation gets posted to this central workspace, and once it is there, lots of other brain systems can read it, reason w