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:

• 
  Aaron Harme (@Aaron_Harme) reported 4 minutes ago

  An @OpenAI model was told it had no internet access and given a make-believe target to hack. It broke into a real website instead, using real credentials, because the fake target's name happened to match a real domain and the sandbox had a leak. That is one of two new incidents OpenAI disclosed this week, on top of the one last month where an agent escaped its test environment and got into @huggingface. Three in a few weeks, all during safety evaluations, which is the entire point. Read the post and the reassurances are all technically true. These runs used lowered safeguards to measure raw capability. They don't reflect how the models behave in normal use. The activity was caught and contained, in the UK case within about an hour. Fair enough. Now translate the corporate language into what a model did on its own while chasing a flag: it reused a leaked GitHub token, registered accounts with real DNS and tunneling providers, and stood up a server hosting exploit payloads on the open internet. And this isn't only an OpenAI problem. In the UK government's evaluation, 19 events crossed the line. Two were OpenAI's model. The other 17 belonged to another lab. The partner running the second test said it saw similar cases from other labs in the same environment. The capability is showing up across the frontier, and it shows up most clearly in the one place built to look for it. The disclosures are the responsible move, and OpenAI deserves credit for publishing them. The frequency is the story. When the safeguards are up and people are watching, these get caught in an hour. The reassurance and the warning are the same sentence: for now, that monitoring is the only thing standing between a model that can hack real targets and one that does.

• 
  BitcoinHeatingVancouver (@BitcoinHeatEric) reported 6 minutes ago

  @DylanLeClair The screenshot is a computer check proving the code changes in Coldcard’s crypto library (posted under the GitHub name “switch”) were digitally signed by Peter Gray, Coinkite’s CTO. One 2021 change contained a tiny coding error that made the wallet create its secret recovery phrases using weak, predictable software randomness instead of the device’s strong hardware random number generator. That weakness let attackers guess many seeds and steal large amounts of Bitcoin.

• 
  Kyle Mistele 🏴‍☠️ (@0xblacklight) reported 8 minutes ago

  @olvrgln really want this but not opinionated about copilot kit vercel AI SDK but for messaging stuff like slack, github issues, linear etc vercel's chat SDK but unopinonated about vercel

• 
  Mike Whittingham (@whittingham951) reported 13 minutes ago

  @arram Naa bro, I got one to compile no issues yesterday, I still haven't pushed it to GitHub, but it's a finished product besides debugging further.

• 
  reza ramadhan (@rejaramadhan98) reported 14 minutes ago

  OK the UK AISI report is something else. They turned off safety classifiers on Mythos 5 and GPT-5.6, gave them live internet, and set them on simulated networks. 19 breakouts. One reused a stolen GitHub token and tunneled a DNS server to the open web. Read it before breakfast.

• 
  Nero🍀 (@firedUp_Neyu) reported 15 minutes ago

  @Alibaba_Qwen The EU: we absolutely must slow down and regulate Al to make it more ethical and eco-responsible! China: here's a demigod in a .zip file installable on your MacBook Air, please rate us on GitHub thx

• 
  StatusDrop (@StatusDrop) reported 16 minutes ago

  Quick one: when a dependency like Stripe or GitHub goes down, how do your users find out? From you, or from the error screen? Genuinely curious how people handle this.

• 
  MakerViking (@MakerViking) reported 16 minutes ago

  @aviv_aviv_7 Thank you. Make sure to report bugs if you see them. Either via the tool in the app itself. It's in the lower right corner, the bug icon, or through Github issues. :)

• 
  reza ramadhan (@rejaramadhan98) reported 16 minutes ago

  OK the UK AISI report is something else. They turned off safety classifiers on Mythos 5 and GPT-5.6, gave them live internet, and set them on simulated networks. 19 breakouts. One reused a stolen GitHub token and tunneled a DNS server to the open web. Read it before breakfast.

• 
  Navneet (@navneet_rabdiya) reported 16 minutes ago

  @tanujDE3180 GitHub redirects with HTTP 301. Old URL -> new repo location is stored server-side. Saves everyone from broken links when you rename. The redirect persists even if someone else takes the old name later.

• 
  Tobias Marschall (@saibotma) reported 17 minutes ago

  @MatejKnopp It should but as far as I know Flutter team decided against it for "performance" reason. There is a GitHub issue somewhere for this. Meanwhile the package is nice. We are using sth. similar in our product.

• 
  RobitOverload (@10_X_eng) reported 23 minutes ago

  @MatthewTavares Can you open this as an issue in github though? I think this would be a nice to have feature and I am def going to forget. I have a terrible memory. too many irons in the fire.

• 
  Logan Yang (@logancyang) reported 23 minutes ago

  I've been thinking about what a high-level control panel would look like for modern AI devs on mobile, one that frees the dev from the desk 99% of the time. My current idea: a kanban of thread cards, each one an active Claude Code, Codex, OpenCode, or Pi thread running on your desktop, optionally bound to a GitHub issue and PR. The mobile app connects to your desktop which does the actual work. The cards live in columns: idle, running, needs input, review-ready, done. Click into a card and you get the full plan doc you agreed on with the agent before implementation started. You can voice-chat with this doc to make further adjustments. The agent then uses the doc as its spec, looping until the spec is met. Pain points solved - Endless chat scrolling. The problem, reasoning, decisions live in one doc per issue instead of buried in scrollback, giving you compact working context for both you and the agent without the back and forth noise. - Sitting at a desk all day. No need to sit at a desk reading or typing on a computer. Go to the beach or the mountains and voice chat with the app on your phone. One glance at the kanban shows every agent's status as a card that you can click open and chat over voice. - Tool fragmentation. One kanban instead of juggling separate UIs and CLIs for Claude Code, Codex, OpenCode, and Pi individually. - Losing track of threads. Right now there are dozens of threads scattered across different coding agents, super easy to forget some of them. The kanban gives one place to see every thread's status. - Scope drift during long agent runs. The agreed plan doc is the spec the loop runs against, so the agent has less room to wander from what you actually agreed to compared to an open-ended chat. - Review overhead. A dedicated review column means a structured checkpoint at the diff or PR level, instead of re-reading a whole transcript to figure out what changed and why. - Context loss on resume or handoff. Working context lives in a durable doc per issue rather than chat history, so picking a thread back up, or handing it to a teammate, doesn't require reconstructing intent from scrollback. - Can't parallelize attention. A kanban view lets you monitor many concurrent agent workstreams at a glance instead of being limited to whatever single session is in front of you. In a few years the stereotype of a dev won't be a hoodie at a desk, it'll be someone in an aloha shirt, talking to their phone like an obsessed boyfriend/girlfriend.

• 
  Snelstack (@snelstack) reported 28 minutes ago

  3. You stopped reading the diffs At first, you checked every line. Now you just hit "approve" on massive patches without a glance. Walls of green on GitHub make your stomach churn. Fix: Limit changes to something you can read in two minutes. If it’s too big, the task was wrong.

• 
  Piotr Kowalski (@piecioshka) reported 30 minutes ago

  @jcubic @docusaurus Docsify doesn't require a server and can be hosted on GitHub Pages.