GitHub Outage Map

Community Discussion

Tips? Frustrations? Share them here. Useful comments include a description of the problem, city and postal code.

Beware of "support numbers" or "recovery" accounts that might be posted below. Make sure to report and downvote those comments. Avoid posting your personal information.

GitHub Issues Reports

Latest outage, problems and issue reports in social media:

• 
  pratik.eth (@eth_ethpratik) reported 10 seconds ago

  @Shahules786 @VibrantLabsAI Hello @Shahules786 , I am trying to report a security vulnerability over the email id provided over GitHub Security.md file but apparently its wasn’t delivered. Please share an alternative email or open the advisory for reporting the issue.

• 
  KS Sreeram (@kssreeram) reported 4 minutes ago

  @Lidinwise @leecronin Given that AI coding is all the rage… What is your hypothesis on why the following is true? AI is unable to create even _one_ open source project that’s good enough to enter the top one-thousand open source projects (say on github), with ZERO involvement of humans from birth of idea. Imagine the prompt being something like “Come up with a great idea for a new open source project and implement it”. AI is unable to do any such thing with zero human involvement. My answer on why: Every project in a top 1000 list is a hit. Every hit is a mini-invention of sorts. It is necessarily “out of distribution” is some way. AI is unable to do this because we don’t know how to solve the problem of invention.

• 
  Yiqing Xu (@xuyiqing) reported 5 minutes ago

  @Faylosophe Certianly. Could you file an issue on the Github page?

• 
  YNWA🐦‍🔥 (@YNWAcrypto) reported 8 minutes ago

  The problem isn’t subtle. GitHub Sponsors has paid out ~$50M total since 2019. core-js: 9 billion downloads, running on half the top 10k sites on earth. Its maintainer was making ~$600/month when he called open source “fundamentally broken.”

• 
  Almog Gavra (@almoggavra) reported 11 minutes ago

  A few other meaningless metrics to optimize for: - I've authored 22% of the RFCs - *** blame marks me responsible for 14% of the LOC (.rs files only) - I've opened 11% of the issues on GitHub - I've generated the most memes on our discord (allegedly)

• 
  Teknium 🪽 (@Teknium) reported 12 minutes ago

  @majoragv Haven't heard of this issue. Do you have an issue on github?

• 
  Gabriel Denys (@gabedenys) reported 12 minutes ago

  @Marcos12345rico I posted a GitHub issue. Assuming you probably want bug reporting mostly there? It's a good tool. Locally I already patched and compiled the app to fix the bug.

• 
  Rahul Verma (@RahulVerma989) reported 12 minutes ago

  @ElitzaVasileva - I have created claude code routines to write blogs for three of my products daily which are driving the traffic from search engines. - You can create a similar workflow to manage your customer support. How 👇🏻 1) Create a feedback menu in the dashboard to create tickets within the platform. One for your users and one for yourself (admin). 2) Create the MCP server and connect it to claude or AI tool that you use. 3) Create a routine so that claude will trigger lets say every morning at 8 AM and go through each ticket and respond. You can also configure webhook to keep it near real time but it might exhaust the usage limit faster. Also include your website github repo in routine so that claude can refer to the codebase to provide accurate instructions. Just instruct claude to not make any edits to your website codebase and respond only when you are not replying for sufficient mount of time (like 3 hours for example) 4) If you are using resend then you can auto create the tickets in the dashboard of the user when the first email is received and after that the ticket will be updated automatically even if you do conversation on email. Like I don't even maintain one of my project LatestModelId as you can see in the screenshot. Claude run each week and update the codebase and I just review and approve the PR. Hope this helps 🙌🏼

• 
  Shubham Sharma | AI & Tech (@editxshub) reported 13 minutes ago

  Paying $19/month for GitHub Copilot? Cascade is free. What you actually get: → Inline completions — not stripped down → Autonomous debugging → Real-time assistance → Command execution Other free alternatives most devs have never tried: → Cline — autonomous VS Code agent (open source) → Aider — terminal-first, built for *** workflows → Continue — local LLMs, data stays on your machine 12 months ago: Copilot was the only serious option. Today: 4 real free alternatives. Most teams paying for Copilot haven't tested any of these. 30 minutes could change a year of costs. Which one are you testing?

• 
  Digita (@digitaworld1) reported 13 minutes ago

  how well a model can fix real bugs in real open-source codebases. It is harder to game than older benchmarks because it uses actual GitHub issues, not synthetic problems. M3 scored 59.0% on SWE-Bench Pro, edging out GPT-5.5 at 58.6% and Google Gemini 3.1 Pro, while sitting just

• 
  naimeh (@naimeh70) reported 15 minutes ago

  @Amir1339216RKT This happens a lot during testnets. Now when I find a minor bug or contract issue, I just drop it publicly on GitHub or tag them directly instead of DMing.

• 
  Conglomerate (@0xconglomerate) reported 16 minutes ago

  Why exactly do VLAs fail? VLAs start w/ LLMs as their brain. Early roboticists (2021-2022) noticed that LLMs trained on internet text had absorbed a large amount of implicit knowledge about the physical world. So they took that best available pretrained brain, observed that actions could be formatted like language tokens, and assumed the transfer would work. But world knowledge encoded in language ≠ physics simulation. There's essentially a data structure mismatch: ▸ LLM pretraining data is discrete, symbolic, and sequential (text). ▸ Physical control is continuous, high-dimensional, and requires split-second feedback. --- ➦ VLAs in the real world, by the numbers: ① They barely work ▸ VLAs start at ~30% success on real robot tasks, it need hundreds of human interventions just to reach ~90% ▸ Best pretrained VLA hit 27.4% task progress on real robots ② VLAs can't generalize outside training ▸ On actions it's never seen, best VLAs score 25-32% task progress (fails when you change the environment) ③ Fine-tuning doesn't help ▸ The more robot-specific, the dumber it gets at everything else (only works on clean, controlled, success-only demos) ④ Too slow for a real robot ▸ OpenVLA runs at 3-5 Hz (physical control needs orders of magnitude faster than that) --- The easiest way to understand how VLAs are actually wrong is thru a real life example. ➦ Let's say you hired a chef who learned everything about cooking by reading, but has never stepped in a kitchen. If you ask them how to cook a steak, they'll tell you the best answer. But if you actually ask them to cook, they'll struggle when you hand them the pan. They'll have a hard time picking up the ingredients. They'll burn the steak. They know everything about cooking, but can't actually cook. --- ➦ Thoughts I want to take back a line I've said before: "Robots can see, but they still can't listen." (referencing to my Silencio piece before) I take it back. Robots can see, listen, even reason now. What they can't do is act in the real world. It's basically an AI chatbot wrapped in a robot body, not a robot that can actually do tasks. No wonder most demos online are scripted. There's a real problem with the brain, and roboticists have been building on the wrong foundation. VLAs are like a trojan horse, they look like the answer but bring a bunch of problems in with them. VLAs only learn through imitation which brings up the data problem. "Enough data" at scale doesn't mean hundreds of demos total. It means hundreds per task, per robot body, per environment. Hundreds again every time any one of those changes. So you've basically got a human-labor bottleneck. To get that data, someone has to physically collect it, either through: ▸ Teleoperation (slow, expensive, needs trained operators) ▸ Kinesthetic teaching (tedious, doesn't scale to complex tasks) ▸ Motion capture (high precision but high setup cost) ▸ Simulation (robots trained in sim often fail in the real world because physics engines aren't accurate enough) And you'd think, okay, maybe someday a company figures out a better way to collect all this. But the problem doesn't stop once you already have the data... Switch to a new robot body and you're collecting data from scratch, because VLAs don't transfer well across embodiments. Move it to a new environment and you're collecting again, since it just overfits to whatever setup it trained on. Give it a new task and yep, collect again, because it can't generalize to actions it hasn't seen. And if you fine-tune it for one thing, you'll probably break another, so now you're collecting data again just to fix what broke. So what was @DrJimFan and @nvidia's answer to this? World Action Models. Instead of building on a language model, you build on a world model: a model that's learned to simulate how the physical world actually behaves. VLA: a language model that learned to output actions WAM: a world simulator that learned to output actions So when you give a VLA a new task, it needs hundreds of demos to learn it. Give a WAM the same task and it simulates it forward first, acts based on that simulation, then adapts with barely any data. This is what NVIDIA did with the first WAM: DreamZero. DreamZero learns by watching the world (any video of anything, not just robot demos). The backbone is a video diffusion model, the same kind of model that generates realistic video. It was pretrained on massive amounts of internet video, so it already learned how the physical world works: how objects fall, how surfaces interact, how motion flows. Doesn't sound like an entirely different approach, right? But NVIDIA looked at it from a different angle. They figured motor actions are shaped a lot like pixels; both are high-dimensional continuous signals. So DreamZero processes them in the same model, at the same time. It predicts the next video frame and the next action together, through the same architecture. So when a robot runs DreamZero, it's literally dreaming a few seconds into the future in video, then reading its own dream to decide what to do next. If the dream looks coherent, the action works. If the dream hallucinates, the action fails. The DreamZero paper dropped last February 2026, and it's been open source on GitHub for anyone to try. Then in March 2026, at GTC, NVIDIA previewed GR00T N2, the direct successor to DreamZero. This is the production version of the WAM architecture, built for humanoid robots at scale And so far, everything's looking promising. GR00T N2 hits a 98% success rate on unseen domestic objects, a 40% jump over GR00T N1 (the VLA), and 2x better generalization than the leading VLAs. NVIDIA swapped robotics' data problem for a compute problem. Instead of collecting more human demos, just simulate more. So yeah, feels like we're finally pointed in the right direction, closer to robots that can actually function in the real world. Excited to see where DreamZero / GR00T N2 goes from here.

• 
  toni (@tonitrades_) reported 23 minutes ago

  @github Capping PRs helps with the queue, but does it fix why reviews pile up in the first place? If reviewers are already stretched thin, limiting submissions might just hide the real problem.

• 
  Xovion Labs (@xovionai) reported 25 minutes ago

  Microsoft just hired AWS to run GitHub. AI demand broke Azure's forecast. From the leaked planning docs: • 2025 Copilot commits: 1B. 2026 projection: 14B • GitHub now does 1.4B commits per month • Copilot error rates peaked at 21% • Planned 10x Azure expansion became 30x in 4 months Owning the data center stops mattering when your own AI floods it. Investors already filed a Copilot disclosure suit.

• 
  Krish Subramanian (@krishnan) reported 26 minutes ago

  Software engineers got automated first. Not because the work was hard. Because it was easy to grade. Everyone blames the missing union. Coders never organized; doctors, lawyers, and electricians did. That is half the story, and the wrong half. Two things get mashed together here: how easy a job is to automate, and who sets the terms when it happens. Take the first. Code is text. The training data sat on GitHub, free. And code grades itself. A compiler and a test suite tell a model in seconds if it was right. That feedback loop is rocket fuel for machine learning, and almost no other job has one. A nurse does not come with a test suite. The result shows. On SWE-bench Verified, a set of real GitHub issues, top agents went from about 20 percent in August 2024 to near 90 percent by early 2026. Human developers score around 67 to 70 percent. The machines have passed us. And the people who built these systems aimed at their own jobs first. The damage is not a prediction. Stanford's payroll data shows employment for developers aged 22 to 25 down nearly 20 percent from its 2022 peak. Now the comfortable read: seniors are fine. Workers over 30 are holding steady. For now, AI writes the code and seniors supply the judgment. "For now" is carrying that whole sentence. Seniors feel safe because the tools write code but cannot yet own messy, ambiguous, system-level problems. That is a line moving up, not a wall. Every benchmark shows models climbing toward harder, multi-file work. Senior judgment is the next rung, not a different ladder. Kill the bottom rung and you kill the pipeline that makes seniors at all. So, the union question, framed properly. A union could not have stopped this. A picket line does not repeal a capability. What it changes is the terms. In 2023 the Writers Guild cut the first real AI deal in any industry. They did not ban the tech. They won this: a studio cannot force you to use AI, AI output cannot take your credit or pay, and the company must give notice first. Engineers won none of that. So the capability landed on the employer's schedule. No warning. No floor. No severance. No seat. Exposure and protection are different levers. Most of us have neither. The juniors already know this. The seniors are next.