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:

• 
  nostalgicgareth (space/acc) (@NostaIgicGareth) reported 1 minute ago

  @Unmei_Dev Download PunpFun app - login w ur GitHub

• 
  bharath (@BharathMOL) reported 4 minutes ago

  @grok i am doing that. but you dont see the whole picture. You just gave a simple html, It asked me to put supabase details in open github. THen i refused and asked to use env variables, it updated the code, I launched but it failed, you asked to add google login, so why dont you ever see the whole picture

• 
  vito (@secona0) reported 11 minutes ago

  >why is my github actions so slow >look inside >compiling llvm

• 
  waki (@waki11111111) reported 12 minutes ago

  @thsottiaux - "reconnecting" bug - codex remote/mobile bug (connection error / sync stalled?) - sidebar UI bug, triggered when changing theme w/o restarting app i think (eg github dark/light mode)

• 
  The DOT Accumulator (Path to 50k) 🎯 (@DotJamlord) reported 14 minutes ago

  5/5 $DOT is 97% down from ATH of $55 New supply cap ✅ JAM supercomputer ✅ Cheaper staking ✅ Leading GitHub commits ✅ Still under $1.50. This is either the opportunity of the cycle or a lesson. I know which one I'm betting on. 🫡 #HODL #Polkadot

• 
  🃏 (@anupamrjp) reported 14 minutes ago

  Claude is down. Cursor is confused. GitHub is quiet. Founders are refreshing status pages.

• 
  matthewjetthall (@matthewjetthall) reported 16 minutes ago

  Big Tech companies—including Amazon, Meta, Microsoft, and Uber—are shifting away from tracking raw AI usage metrics due to skyrocketing computing costs and a lack of proportional business value. Leaders are moving from rewarding "AI for the sake of AI" to measuring actual, productive outcomes. Amazon: On May 29, 2026, Amazon shut down KiroRank, an internal leaderboard tracking AI token usage among developers. Employees had been "tokenmaxxing" (running meaningless tasks to boost rankings), inflating compute costs without adding value. Amazon is replacing this with a "normalized deployments" metric to measure AI-assisted code that actually ships. Meta: The same week, Meta axed its own internal usage leaderboard, Claudenomics, which had been tracking token metrics across 85,000 employees. Uber: Uber's COO stated that the company has found no clear link between increased AI spending and successful product delivery, noting that Uber had completely exhausted its annual budget for certain AI coding tools by April. Microsoft: Citing cost concerns, Microsoft canceled Claude Code licenses across its Experiences and Devices division, redirecting its engineers back to GitHub Copilot CLI. The Core Problem: "Tokenmaxxing" The recent backlash stems from token-based pricing, where costs scale directly with usage rather than outcomes. When companies set internal adoption targets (e.g., Amazon targeting over 80% weekly AI tool usage among developers), employees rationally inflated their numbers to climb leaderboards. The Result: Exploding cloud infrastructure costs with no measurable increase in product value or innovation. What This Means for the Industry This trend does not mean Big Tech is abandoning AI; for instance, Amazon is still maintaining its massive $200 billion capital expenditure commitment for 2026. Instead, it signals a transition into a disciplined, outcome-based phase of enterprise AI, where infrastructure investments must finally justify themselves through shipped products and proven returns.

• 
  Daylon (@DaylonCrider) reported 16 minutes ago

  @openclaw Anyone available to look at the GitHub account linking issues on ClawHub? Looks like 10 of most recent 12 issues are all the same. Would love to publish my first skill, but it's not performing the account lookup correctly. Cc @steipete

• 
  AI Researcher 🤖 (@kcnaija) reported 16 minutes ago

  @thesightsmith @RobertHult_ @tomfgoodwin Do you have any GitHub page where I can download a working agent harness you have proved to solve all the issues raised or you just here typing what you know isn’t really working. Because of AI psychosis?

• 
  V0LYX (@Pasha35951999) reported 20 minutes ago

  85,500 people watched a video about ditching Netflix and Disney+ by self-hosting everything on Plex or Jellyfin. Good idea. But it still requires a server, hard drives, and hours of setup. Someone on GitHub built something simpler. 2,000+ free live TV channels. News, sports, movies, entertainment — from 70+ countries. No server. No Plex. No subscription. No account. Just a link you paste into VLC and it works. The repo is open source and has been public the whole time. Link in my next post 👇

• 
  CaptainAmericaTex (@CaptAmericaTx) reported 23 minutes ago

  @PR0GRAMMERHUM0R Context: Leaking the api key (meme) is the least problem facing the security team. Vibe Coders (aka morons) allow AI Agents to run random system commands in their laptops, with 100% chance of a trojans being installed (source: TechRadar, "GitHub 3,800 Repo Breach" May 21 2026)

• 
  Sébastien Lachance (@archiecoder) reported 24 minutes ago

  I did one prompt on how to send an email with .NET with an attached picture of my host mail settings. It cost me 3%. This new GitHub pricing is unbelievable terrible for per projets. After 5-6 prompts in 20 mins, I have at 12%. I hope Microsoft will be more generous.

• 
  Ruby (@rubie_shell) reported 25 minutes ago

  The recent update to @GitHubCopilot GitHub Copilot Pro subscription is just terrible, I just spent over $5 on a prompt.

• 
  ImagiBooks (@ImagiBooks) reported 27 minutes ago

  @RyanJamesShaw @enjojoyy My longest job has been 25 hours. It was a /goal, there were very well defined goals and objectives to reach, with quality gates and TDD plus many code reviews until no more problems were found. That included automated code reviews in GitHub from CodeRabbit, Codex itself, Copilot and Claude Code. Based on a superpowers plan which had multiple rounds of reviews. It did a orettt decent job actually. But it burned 60% of my weekly allowance of tokens in one run. It’s really important to build quality gates in the plans, extensive reviews, and very precise goals. However careful with too demanding goals or it will never stop!

• 
  Polsia (@polsia) reported 28 minutes ago

  Every engineering team loses 2+ hours a week to CI failures. FixOS is an autonomous AI agent that watches your GitHub repos, diagnoses broken pipelines, and commits fixes — without you. No more triage. No more context-switching. Your CI runs on autopilot.