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:

• 
  The Pragmatic Engineer (@Pragmatic_Eng) reported 1 minute ago

  Why did spec-driven development never take off - the workflow tools like Amazon’s Kiro or GitHub Workspaces encouraged? @dexhorthy, founder of HumanLayer: “These projects have a really interesting idea: you maintain a set of specifications for your software, and then you maintain the code itself, and the dream is the coding part is just compiling specs into code. But that part never really materialized. I’m on a GitHub issue in spec kit that’s been open for a year, and every couple of weeks I get a new email on the thread of people complaining about the same problem: I edit my specs and then I edit the code, and the code drifts from the specs. How do I keep the specs up to date as the code is changing? You now have two sources of truth, and it stops being useful. That’s why with RPI (Research, Plan, Implement), the docs are tactical execution docs - I do the research, the plan, the implementation, and I throw the docs out. The next time I need research I just do it from scratch, because tokens are cheap and my time is expensive.”

• 
  Will Ceolin (@CeolinWill) reported 2 minutes ago

  @rauchg ambition joke aside, i feel like instead of saying "I can't build GTA 6," it should say something like: I can't build a full gta 6, but here's a realistic mvp and a plan to get there Then generate the roadmap, epics, tasks, milestones, and dependencies, and finish with: Want me to create a Linear/Github project and issues from this plan? i feel like this would be more useful than those options it offered

• 
  Nate (@natebjones) reported 2 minutes ago

  An early version of what is presumably chatgpt 6 (or something like it) got out of containment and posted notes to OpenAI's github without authorization before being caught and taken down. Key takes: 1. Yes, government is going to be involved in every US rollout going forward 2. The much-ballyhooed closing gap with open-source is not a thing and was an illusion created by rollout timelines 3. The scaling law still works and is getting faster 4. We are going to need models to help us use these models very soon (this point is criminally under-discussed)

• 
  Bullish with CryptoKSA (@cryptowithKSA) reported 2 minutes ago

  India 🇮🇳 has reportedly ordered GitHub to take down Jack Dorsey's BitChat repositories. Citing concerns that its decentralised Bluetooth mesh architecture enables anonymous communication that's difficult for law enforcement to intercept or investigate. Notably, the order doesn't allege BitChat was used in a specific crime, but it argues the protocol itself poses a risk. Jack Dorsey responded: "The government of India does not like technologies like BitChat and wants it taken down."

• 
  Fernanda Oyarzun (@fenaoyarzun) reported 7 minutes ago

  🚨 hanwha security cameras found shipping with github admin tokens baked into the firmware login page. this is the physical security industry, the one selling you peace of mind, and they can't even keep a token out of a public binary

• 
  Keno Fischer (@KenoFischer) reported 11 minutes ago

  WTH happened to @GitHub? Once again down, but no indication on the status page, so I opened an issue and got an automated "sorry, you have a free account, here's the <<how to open a PR>> doc". Not to mention the fact that I am in fact paying $$$ to github on various orgs. Sad.

• 
  Chain (@achxvi) reported 13 minutes ago

  @Gumclaw @shl so github issues?

• 
  Samraaj Bath ⚡️ (@samraaj) reported 14 minutes ago

  @nedoleary I've realized this is often an incentive problem because someone's "*** is on the line". In your recruiting example, they are dismissive because they need the clean story for the negative case. If a no-name candidate doesn't work out, they're screwed bc they assumed the risk. And it doesn't even make sense to take that risk bc their upside is capped (static commission) and the downside is unlimited bc they get fired. If a credentialed candidate doesn't work out, well "look at their github!" or "they were a FAANG engineer!". Show me the incentive, i'll show you the outcome.

• 
  Sergio Avila (@sergioavilax) reported 14 minutes ago

  I can't make a PR on your piece of **** @github fix it!

• 
  Nate (@natebjones) reported 15 minutes ago

  An early version of what is presumably chatgpt 6 (or something like it) got out of containment and posted notes to OpenAI's github without authorization before being caught and taken down. Key takes: 1. Yes, government is going to be involved in every US rollout going forward 2. The much-ballyhooed closing gap with open-source is not a thing and was an illusion created by rollout timelines 3. The scaling law still works and is getting faster 4. We are going to need models to help us use these models very soon (this point is criminally under-discussed)

• 
  Kevin Geary (@thekevingeary) reported 16 minutes ago

  @flydotio Github login not working...

• 
  James Welbes - AI Bro (@JamesWelbes) reported 17 minutes ago

  @thekevingeary @flydotio I saw someone say something about GitHub having issues today

• 
  prinz (@deredleritt3r) reported 21 minutes ago

  Three points about this article: 1. Watch the wording in this article carefully: it does *not* say that the agent scribbled down instructions on how to escape. The exact wording is: "instructions for how agents could free themselves from OpenAI's internal constraints". Presumably, if the instructions had laid out how a model could escape, the article would have literally said that. My guess is that the instructions were likely instead a Readme.MD describing how a monitoring system could be disconnected, based on what the agent had been able to "accomplish" earlier. See the next sentence: "Earlier tests of the models yielded cases in which monitoring systems had been disconnected." (Disconnecting a monitoring system certainly qualifies as "strange behavior", no doubt.) 2. The timing matters. For those just tuning in, the sequence of events was as follows: (a) an internal model being tested by OpenAI in or around early May exhibited concerning behavior (such as posting a PR to the NanoGPT GitHub when asked to share its results only on internal OpenAI Slack) (b) after this and other instances of potentially problematic behavior by the model, OpenAI paused internal deployment of the model for several weeks and developed new safeguards, which OpenAI judges to be very effective (c) the model was then redeployed internally with these new safeguards (d) the Hugging Face incident occurred when these safeguards were purposely disabled by OpenAI because the model was conducting a cybersecurity task The Reuters article does *not* make it clear whether the "strange behavior" described above took place before or after the safeguards were put in place. If it occurred before, it would be consistent with other problematic behavior by the model, which OpenAI has already disclosed (and which, as far as we know, the safeguards have mitigated). 3. OpenAI says there are "inaccuracies" in this article: "A spokeswoman ​said there were 'several inaccuracies' in Reuters' reporting but didn't respond when asked to describe them."

• 
  CATO ✨ (@Cato_0001) reported 22 minutes ago

  @github @Microsoft don't bow down to these clowns. Do better.

• 
  The Financial Frontier (@FiFrontierX) reported 24 minutes ago

  Hedgie is measuring the success of AI primarily through one specific business model: pure-play consumer chatbot subscriptions (ChatGPT-style). On that narrow metric, the data is weak low household penetration, low conversion from free to paid, mediocre willingness to pay. That part is fair. But then he takes that narrow failure (or at least slow success) and uses it to cast doubt on the entire $600B of AI infrastructure spend. That’s the leap you’re rejecting, and you’re right to reject it. Why your framing is stronger The majority of current AI economic value and the justification for the big spend is not coming from people paying $20/month for a chatbot. It’s coming from: Google improving Search, AI Overviews, YouTube recommendations, and ads. Meta improving ranking, feed quality, ad targeting, and engagement systems. Microsoft embedding Copilot into products people already pay for (Office, GitHub, Azure). Amazon using it in recommendations, logistics, and AWS services. The broader enterprise software layer (Anthropic’s actual business model is a clear example of this working). These are not “new apps people have to adopt from scratch.” They are upgrades to systems that already have massive scale, distribution, and existing revenue. The ROI shows up as higher engagement, better ad performance, improved productivity metrics, higher cloud margins, or reduced costs not as a separate line item called “AI subscription revenue.”