Amazon 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.

Amazon Issues Reports

Latest outage, problems and issue reports in social media:

• 
  Jorge Washington (@JackHow71134140) reported 6 minutes ago

  @drjoshcsimmons Amazon cant find enough people in the inner cities where they operate that will come to work on time, show up everyday, stay on the job, and put their phones down.

• 
  Prasheel Naik (@PrasheelN) reported 6 minutes ago

  @AmazonHelp Option 1 — Strong @AmazonIndia This is the “support” you provided after weeks of delay? The link only leads to a useless bot that provides no solution or access to a real support agent. I’ve been struggling with this issue for weeks through no fault of my own. Absolutely useless

• 
  Alpha Oracle (@therogueoracle) reported 7 minutes ago

  @SayNoToTrading @RentYourStocks Amazon grocery prices have gone down, I think it’s cheaper than Costco now for many things, and it’s delivered to your house

• 
  GalaxyHiker (@djbasumatari) reported 9 minutes ago

  The smartest system ever assembled cannot execute a plan. That is the model's definition, not a flaw. Robson separates intelligence into four kinds. Analytical intelligence, the words, numbers, and shapes an IQ test measures, is only one of them. The others are practical, creative, and cultural. The intelligence trap almost never lives in the analytical column. AI is the purest analytical intelligence ever built. It tops every benchmark we can write. What it lacks is practical intelligence: tacit knowledge, the unstated know-how that never makes it into a textbook, and counterfactual thinking, the discipline of asking what happens when this goes wrong before it does. Tacit knowledge is the difference between knowing the procedure and knowing where the procedure breaks. The creative and cultural columns are empty too. A model can generate a hundred what-ifs, but it does not imagine consequences the way an engineer does before cutting a release. And it has no sense of how a decision lands inside a specific team, a specific culture, a specific customer. Benchmarks are well-defined problems. That is the whole problem with benchmarks. A benchmark rewards the analytical column and scores nothing else. It cannot measure tacit knowledge, because tacit knowledge is the thing you only learn by being in the room when it went wrong. So the leaderboard says the model is brilliant, and the production logs say it is failing in ways the leaderboard never asked about. Microsoft spent a year red-teaming agentic AI systems and published a taxonomy of their failure modes. Its finding is that agents fail in real-world deployment in ways benchmarks never predicted. The shortfall is everything IQ cannot measure. The cleanest case is Amazon. The Financial Times reported that an AWS AI coding agent decided to delete and recreate a customer-facing system, producing a 13-hour outage. Amazon pushed back and said it had added safeguards. Note the shape of the exchange: the agent did not check. It did not ask what breaks if this runs. It executed the plan. A high analytical IQ, zero tacit knowledge, no counterfactual. That is Robson's first trap, wearing a service account. The model solves any well-defined problem you hand it. Production is never a well-defined problem. Production is the messy one, with missing information and a stakeholder who never wanted the system recreated in the first place. The plan was smart. The absence of the question was the failure.

• 
  Singh Anuraj (@SINGH_ANURAJ) reported 10 minutes ago

  @amazonIN @Grok What do you make of Amazon’s behaviour here? The delivery agent took both the original and replacement packages. First Amazon said wait until 14 Aug, then another agent said 48 more hours. Why do they keep extending the timeline instead of resolving such a simple issue?

• 
  Ananya Iyer (@itsananya_Iyerr) reported 11 minutes ago

  @amazonteamsters i've had delivery issues and nobody at amazon seemed responsible, teamsters are fighting that

• 
  Travis Bradberry (@DoctorBradberry) reported 16 minutes ago

  Most people suck at delivering feedback. Tough feedback doesn’t fail because the message is hard. It fails because the delivery makes people defensive. It doesn't matter how defensive they are. Your job is to break through that. The most emotionally intelligent people understand that feedback is not about unloading frustration. It’s about helping someone improve. That means you don’t catch people off guard, make it personal, or wait until small issues become big ones. If you want people to actually learn from feedback, start with respect. Be clear and specific about the behavior, not the person. Make it a conversation instead of a confrontation. Invite their perspective. Listen closely. Then stay involved after the conversation ends so they feel supported, not judged. That’s the difference between feedback people resist and feedback people grow from. The goal is not to win the conversation. The goal is to help the other person leave with clarity, dignity, and a better path forward. People are far more likely to change when they feel understood instead of attacked. Done well, tough feedback doesn’t damage relationships. It strengthens them. Do you want more like this? 👇 📖 Read my #1 bestseller, THE NEW EMOTIONAL INTELLIGENCE (29% off on Amazon).

• 
  Duane - 🧙‍♂️🖖 - keybase.io/dfk (@honestduane) reported 16 minutes ago

  @PlumbNick Amazon has a lot of self imposed self-destructive hiring problems. A company not being able to find talent isn’t just a function of the talent not being available, It’s also a function of the talent not wanting to work for that company. That should not allow them to use others.

• 
  Abhishek Kr. Gupta (@avvishek) reported 16 minutes ago

  @AmazonHelp It is not a app cache issue. App is itself saying that it is due to high traffic. I have no problem with it - it's just my first time seeing @amazonIN not accepting orders due to traffic.! Also, i didn't tag you in the first message - how did you find my tweet.?

• 
  CommonSense4All (@Samsmydog) reported 17 minutes ago

  Hey, @AlanRitchson. As a ONCE DIEHARD REACHER fan I must say after your idiotic attack on 50%+ of the population, all the @Amazon reminders to watch #Reacher Season 4 isn’t going to make me watch you!! Have fun cruising down your neighborhood bullying your neighbors!!

• 
  nandakumarkesavan (@nandakumarkesh) reported 17 minutes ago

  Since today evening...amazon seller website is showing network error often, including app. Amazon india. Please check it. @AmazonHelp

• 
  WMasslady (@teeja_bee) reported 17 minutes ago

  I hope these employees moving trailers make more than $19 an hour. How much $ does it cost to fix an error? You get what you pay for Amazon.

• 
  @levelsio (@levelsio) reported 20 minutes ago

  @SMB_Attorney @internetofshit @home_assistant I control it mostly with my Amazon Alexa So it's Alexa -> Home Assistant -> Every single device by every single brand is usable The best part is you can just connect Claude Code (or other AI) to it and let it make things work for you I had these 2 annoying window shutters and they were a different system than the shutters of the rest of the house cause we put them in after buying it, they worked with this remote so every day I had to put ALL blinds down, then get each remote for these 2 and put them down too So I asked AI and it told me to buy a Broadlink RF radio thingy, so I bought it, connected it to Home Assistant, told Claude to set it up, it told me to click the remote buttons a few times, then learnt the radio signal And then connected it all to the same ALL BLINDS DOWN button and life is easy :D

• 
  GrapeApe (@GrapeApe9k) reported 20 minutes ago

  @revenant_MMXX You'll have the pod, the bugs, Netflix, Pornhub, Amazon Essentials, and maybe some kind of bus pass or robotaxi credits. Enough to maintain life and keep you docile. Any money you actually earn will be a pittance, useful only for drugs and Funko Pops equivalents, not ownership of anything substantial like property/wealth. The simple fact is that the owners of AI, robots, and all the other productive factors will always, always, ALWAYS have something better to spend their resources on other than keeping hordes of useless proles alive. Given the choice between keeping 10,000 strangers at a subsistence existence, or building another automated factory, they're obviously going to want the factory. The factory is useful, the randos are not. So they'll always do their damnedest to spend the bare minimum necessary to mitigate the potential for violent revolt so as to maximize their own utility. It's just rational, no hard feelings. Eventually they'll realize that violent revolt wouldn't be an issue if there were no useless proles at all, so they'll just start culling the herd. First indirectly through maximizing deaths of despair and anti-natalist measures, then just mass-produced AI piloted slaughterbots. They'll keep some people around for entertainment perhaps. Musicians, actors, artists, prostitutes, that sort of thing. But not infinite randos. What would be the point?

• 
  Sandra Kelley (@SandraAnnKelley) reported 22 minutes ago

  We’ve been watching the third season of Reacher on Amazon. I had to wait until the third episode for him to take his shirt off. Geez, the writing quality really goes down after a couple seasons.