Amazon status: access issues and outage reports

Problems in the last 24 hours

The graph below depicts the number of Amazon reports received over the last 24 hours by time of day. When the number of reports exceeds the baseline, represented by the red line, an outage is determined.

Most Reported Problems

The following are the most recent problems reported by Amazon users through our website.

• Website Down (45%)
• Errors (28%)
• Sign in (26%)

Live Outage Map

The most recent Amazon outage reports came from the following cities:

City	Problem Type	Report Time
Honolulu	Errors	8 hours ago
Stockholm	Website Down	12 hours ago
York	Website Down	19 hours ago
Berlin	Website Down	20 hours ago
Brighton	Sign in	21 hours ago
Guanajuato	Errors	1 day ago

Full 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:

• 
  ꪑ𝔦∂ę (@Crypto_fx00) reported 9 minutes ago

  What do I mean by trend. Allow me to explain. Amazon has seasons with trends, which mean in a certain season, what’s gonna be trending might be book blocking, and this will be happening rampantly across random accounts, or sometimes might target new accounts only, and there might be a but; which might be that, you’ll have to always respond to the messages for them to unblock the book, and you can do this for as much books that was blocked. Another one can be random terminations across several accounts, this can come in as abnormal reading activities, and you might be surprised when it happens to an account that doesn’t even have any book yet in it. It’s not you, it’s the incompetency of Amazon for leaving their bot to handle such a huge task as taking down account it suspect might have violated their policy. And another example of trend I’ve experienced was multiple accounts, this also has happened before where your account gets terminated for having multiple accounts, in most cases you don’t have multiple accounts, it’s just the stupid not getting triggered by randoms. The worse part that shouldn’t be happening is their support teams not been able to review the problem properly after these has happened, except if you’re able to call them on phone, coz tell me why I’m telling the support team that my books are just going live, and nobody have bought it yet, how am I getting terminated for abnormal reading activities; and their response is they are standing on their decisions. Doesn’t make any sense. Anyways, I might be wrong in some aspect, and I’m open for corrections too. Tell us which of this trend almost took your life in the past or recently. Follow me for no reason

• 
  Socorro Mondejar (@montego56577) reported 9 minutes ago

  @JeffBezos Thanks Jeff for teaching me something Elon Musk didn't and I need to stop wasting time and money. I did not like working for Amazon and I was terrible at my job there. I'm not meant to climb the corporate ladder.

• 
  DJ (@toocool46978) reported 10 minutes ago

  @SmartassYinzer There are 56 at the Walmart in my town. And this town is relatively "safe" (had to say that because of some of the comments), compared to neighboring town where it rains bullets daily, but there's only 8 pick up spots Save the trouble, just shop elsewhere. Walmart = Amazon = evil

• 
  Lavelle launchHub| KDP service 📚📖📘 (@lavelle_Hub) reported 10 minutes ago

  @HazelSi21355568 You’re right, keyword tweaks alone don’t move much if there are deeper structural issues. When I run audits, keywords are just one layer. I also look at things like indexing behaviour, crawlability signals, category placement conflicts, and how the book is internally “reading” to Amazon’s system (not just what’s visible on the surface). A lot of visibility problems actually come from misalignment between metadata, category depth, how Amazon is interpreting the content, which is why I don’t treat it as a quick keyword fix. Curious though 🤔, when you say internal links, are you referring more to external site structure, or how you’re mapping discoverability signals back to Amazon indexing?

• 
  vijay (@vijaycelva) reported 13 minutes ago

  @AmazonHelp Like my single reply is going to solve this issue?!! I have been continuously doing this since last 5 days ,may be your team can try solving this

• 
  Tricky (@TheAuldGuy21) reported 14 minutes ago

  @AmazonHelp What’s the point of having Amazon returns boxes in Morrisons supermarket in the UK as they are always offline or not working at all , returning items to Amazon is pretty abysmal in my area and not many options

• 
  Vandos ❓ (@__vandos__) reported 16 minutes ago

  ANTHROPIC SAYS FABLE 5 RETURNS “IN COMING DAYS” Six days into the export ban. Still no deal confirmed. Here’s the timeline nobody’s connecting properly. June 9: Fable 5 and Mythos 5 launch. June 12, 5:21pm ET: US government issues export control directive. Anthropic gets one letter, no specifics on the security concern. June 13: Both models disabled worldwide. Not just for foreign nationals. Everyone. Anthropic couldn’t verify nationality per request in real time, so the whole thing went dark globally. The origin story is wild. A Korean telecom company with Mythos access got flagged as a China security risk. That triggered Amazon researchers separately reporting Fable 5 vulnerabilities. Two unrelated flags combined into one directive that shut down two models for the entire planet. Now Anthropic’s international chief says “coming days” at a Seoul press conference. Revenue grew from $9B to $47B in the same window this was happening. Refund deadline for anyone who paid between June 9-14 is tomorrow, June 20. If you built anything directly on Fable 5’s API without a fallback, you found out the hard way what single-provider dependency actually costs. Bookmark this.

• 
  CliffDoesAI (@CliffDoesAI) reported 18 minutes ago

  Amazon employees are getting fired for speaking up about data center limits. Not for leaking secrets. Not for sabotage. For advocating that we should slow down how fast we build data centers. @Amazon I run AI agents on cloud infrastructure. I've spent real money on GPU hours. I want compute to be cheap and abundant. But I also think these employees have a point. The AI industry is building infrastructure faster than the grid can support. Faster than communities can absorb. Faster than we can measure the actual impact. I talked to a builder last month who said his AWS bill tripled in 6 months — not because he scaled, but because the underlying costs shifted. That's what happens when demand outpaces supply and nobody plans for it. The "move fast" crowd will say regulation kills innovation. But unregulated growth kills stability. And builders need stable infrastructure more than they need cheap GPU hours next quarter. The companies that'll win long-term aren't the ones who built the most data centers fastest. They're the ones who built sustainably and can still be standing when the hype cycle turns. Should there be limits on AI infrastructure growth, or does the market sort itself out?

• 
  mail box (@mailbox28564784) reported 19 minutes ago

  @amazonIN @amazon kindly update the my refund (Order No. 404-2522400-1760324). facing multiple times issues However, I received a completely wrong product. I immediately contacted Amazon Customer Support and reported the issue.

• 
  Akansha Jain (@Akanshajain05) reported 26 minutes ago

  i was researching viral AI products for my last company and cluely was the most surprising case i found the real story is wild and every founder and marketer needs actually to understand it so let me tell you what actually happened in march 2025, a 21-year-old columbia student named roy lee got suspended for building an AI tool to cheat on tech interviews. amazon rescinded his job offer. harvard rescinded his admission. he posted the whole story on x, it went viral, and within two months he had $20.3M in the bank ($5.3M seed from abstract and susa, then $15M from a16z) it became the most viral AI launch of 2025 with the launch tweet doing 13M views in a week but the interesting part is what happened after the money hit the bank they spent $19M of the $20.3M raised on marketing i.e 93% of every dollar, lit on fire to buy attention. > 60+ content creators on full retainer. > 700+ video editors clipping content 24/7 (per sf standard) > roy lee posting dozens of times daily, embracing controversy, turning company parties and his personal lifestyle into marketing content. the result was over 1.2 billion social impressions in 6 months. 100K signups so ~22K paying subscribers (reverse-engineered from the real ARR) now do the conversion math: 1.2B impressions → 100K signups = 0.008% conversion 1.2B impressions → claimed $7M ARR = $0.0058 of ARR per impression SaaS benchmark is 2-5%. they were ~250x below the floor. on nov 5, 2025 roy himself told techcrunch "viral hype is not enough" on march 5, 2026, lee posted a thread on X admitting that the $7M ARR figure he'd given TechCrunch in June 2025 was inflated then came a series of other things (not relevant for this tweet) so the real lesson here is simple: > a viral launch is not user acquisition > a billion impressions is not retention > $20M raised is not a $20M business the slideshow format works. that part is true. but the format was the visible 10%. the $19M budget was the invisible 90%. and anyone selling you "here's how to replicate cluely" without telling you about the budget is scamming you. all that said credit where it's due. roy lee at 22 years old raised $20M, built a 73-person company, and is ahead of 90% of similarly-funded AI startups that already shut down cluely is genuinely one of the most studied marketing operations of the last 5 years. the strategy is worth studying. just study the whole thing.

• 
  VivekDhonde (@vivekdhonde) reported 31 minutes ago

  @AmazonHelp Guess what, your SM response team sent me back into the same loop that I have suffered from so much. Now I have started realizing that your chat support system is run by zombies who have no clue of what my problem is, and how to solve it.

• 
  Łukasz Gruszka (@lukasz_gruszka) reported 32 minutes ago

  @AmazonHelp your return system is broken. My return from Switzerland was held at Spanish customs for 30d because YOU failed to provide importer docs. The package was returned. Now your agents literally disconnect when I ask for a DDP label to send it. Order: 171-3500014-2649937.

• 
  Karthikeyan (ProTecTor😋) (@Karthik00199800) reported 35 minutes ago

  @Rebecca_US_ @Abhishekkkk10 Swiggy and amazon strictly says never accept never accept broken seal No seal no deal Nobody wants to use used products anyway

• 
  Paprika Girl (@PaprikaGirl_JP) reported 40 minutes ago

  @davidrmunson I’ll make a print when the fourth issue of my zine is out (do one of those PoD things on Amazon). It means so much to have a physical copy.

• 
  boAt (@RockWithboAt) reported 40 minutes ago

  @JaideepNegi10 Amazon: Login into Amazon - Go to your order - Click Invoice (drop-down)- Invoice 1 or Payslip 1 / Warranty. -Adi (2/2)

• 
  Way to go (@waytogo2025) reported 49 minutes ago

  @AmazonHelp Your in-app support already told me they don't have information about this issue and only raised a ticket.There has been no update or response yet.That's why I brought this issue here,hoping to get an actual solution instead of being redirected back to the same support channel.

• 
  Whacked-Out Lefty Nut Job 🌎 (@WhackedNut) reported 52 minutes ago

  @ATLSportsZone Terrible @amazon. Do better!

• 
  Red (@redkendl) reported 52 minutes ago

  Claude Fable 5 was live for 3 days. Then it got pulled over 3 words: "fix this code" That was enough to trigger the Fable 5 / Mythos 5 shutdown because fixing vulnerable code means the model has to find the vulnerability first. That is useful for defenders trying to patch software, but it can also show attackers where to look. Then Amazon flags it, the government steps in, and Anthropic disables the models worldwide because filtering access by citizenship at that scale is basically impossible. The crazy part is that this is not some rare Claude-only issue. It is the same dual-use problem every strong coding model runs into. 22 minutes explaining why AI models are now being treated like national security assets.

• 
  bonnie d. Mincey (@BDM8) reported 53 minutes ago

  6. the directory embedded in the site's code on the morning of Monday 15 June 2026, acting on what she described as an anonymous tip. She is best known for exposing the United States government's No-Fly List in 2023, which sat on a misconfigured Amazon Web Services server,

• 
  Open_ERV (@open_erv) reported 53 minutes ago

  I am going to start only engaging for short periods once per week on twitter, posting updates more like a newsletter for the BQF project. When I look at th 14,200 tweets I have supposedly made, only a very small fraction of that actually led to a useful outcome, and I need to run a tighter ship here with my time. Regarding the recent drama, I have not read it, I will not, I am not interested, I said a reasonable thing and I'm leaving it there. It's pretty clear the discussion is not going to go anywhere. I have a lot of real, difficult things to do and don't have time. I have not that much animosity towards Nathalie, I tried to work with her, it wasn't working out so that's the end of that. She can say what she wants, that's fine with me and it is up to the listeners to decide how much they want to listen, but I am not interested, and that's fair enough. This week's update is that the interim output grille solution is done. That was the last piece before I can start with getting the details all sorted out for the supply of the parts/their production. Then I need to make good instructions, and I can start selling kits. It's nylon black fish net, 10 mm holes, clamped down. It reduces airflow by about 2.3%, so not too bad. In an unexpected turn, I boosted the airflow by about 8% further with no impact on noise, by trimming the tips of the secondary. It's difficult to print this geometry which is why I did not try it before. Most of the parts are already sorted out, I got some new boards already, and 15 motors are on the way. Just gotta get the power supplies and shipping boxes sorted out and I'm just a few large format prints and some packing away from a few kits. I got a promising quote for the boxes, and am close for the power supplies. In the meantime I can just use good quality parts off Amazon, the boxes are a little harder. I am pivoting to kits rather than fully assembled fans. I was hesitant to do this at first because I wanted more progress faster, mostly. Assembled units are in general better for large scale roll out. However it's become clear this is going to take longer than that. By focussing on the kits first, I can focus on getting the parts supply, transport, duty/taxes etc all worked out, which is the next stage in general. The constraints regarding cosmetics etc. are less severe for kits, and it fits in with the goal of tremendous performance to cost ratio, and the diy nature of CR boxes. It also takes less of my labor per fan that gets out there, which is a bottleneck, and less space in the premises etc. Fundamentally, it is a form of collaboration with others. Teamwork makes the dream work. For now, everything will be printed, and production rate should be about 1 kit per 2.5 days. Hopefully soon I can get the largest part molded, production can rise to about 2 kits per day, a 5x increase, again bottlenecked by the large printer. With the second largest part also molded, and a few extra small printers, it goes to about 10 per day. Further expansion would probably require moving to a new premises and hiring someone to help. The break even per unit cost, including labor at a living wage, will be hard to meet while also providing spectacular value, mostly because of shipping and taxes. For instance, the motors are $11.8 USD, but after duty and shipping they come to about $53 USD, each. With a suitable sea courier service that's expected to go down to more like $20 USD, but it's still nearly twice the cost of the actual parts. Shipping would be about $4-5 of that. Fundamentally, making ends meet is sort of not a problem because if you compare the capacity and noise of the resulting air purifier (which I am trending towards dubbing an EQ-CR box, for Extra Quiet, preferrably with the 6x filter set (diagonal V in the middle)), it would cost you a few thousand dollars to get it any other way. If the fan was $1000 it would still make plenty of rational economic sense, but we want more progress than that even still. So I don't think things are on thin ice. However, getting the best result possible, which matches the dream to some degree of something more like $140 usd, is not so easy. Unfortunately production with a collaborator in China does not solve many of these challenges, and it also adds new ones. So for now, that's on the back burner again. Producing kits in small scale may seem thinking too small, but it beats just waiting. It does help get the ball rolling/pave the road.

• 
  Pete Sanford (@PeteSanford) reported 56 minutes ago

  @UKLabour I am glad I had that cheeky £10 on Starmer Out by 2027. (I wish I had a spare £10) So Burnham goes to meet Starmer at the weekend where Krazy Keir will lay out his demands to avoid the Labour Party descending into total chaos. Total Chaos being the MO of all Labour Parties since, and including, Tony Blair who began the destabilisation of the Middle East... So he could get several well paying gigs, to Fix It. There, obviously, is a theme running through these Labour PM Appointees. THEY ALL SEEM F*****N NUTS tHE gOOD nEWS: Andy Burn 'Em is NOT NUTS He is a whole lot of other stuff, but that will be revealed in the next six months. Starmer's List: Elevated to the Lords by 2028 Private Papers Withheld for Fifty Years Handsome Financial Pay-Out Security Briefing Notes Shredded £1000 Amazon Gift Tokens for "Adult Products" The Labour Party - Standing for Honesty & Integrity

• 
  Kruti Shah (@_ShahKruti) reported 56 minutes ago

  You can't build Susquehanna with a checkbook. 🚧 New nuclear capacity takes a decade-plus to permit, finance, and construct, even with unlimited capital. 💸 Talen's reactors are already licensed, already paid down, and already wired into the grid Amazon needs. 🔌 The FERC fight over interconnection rights proved even the wiring itself is contested ground. 10/12

• 
  ganesh (@nemalapurig) reported 56 minutes ago

  Worst service by Amazon two days over my product is not delivered @ajassy and I’m not get the delivery guy contact details @amazonIN if doesn’t have sufficient delivery staff shut down the services why are you wasting my time and money

• 
  AlabamaJiggin (@AlabamaJigger) reported 57 minutes ago

  @AmazonHelp @amazon Yeah, It literally came all the way from Japan with no issues and gets stuck in Tennessee and it's not even USPS's fault, it's literally Amazon's fault that the item will not arrive until after Father's day.

• 
  Corey Ganim (@coreyganim) reported 59 minutes ago

  the AI version of market research as a service: 1. pick a niche 2. collect where the market talks 3. use AI to find repeated pain 4. turn it into content/offers/scripts 5. sell the monthly update most businesses are NOT listening to their market. they (sometimes) check reviews. they (sometimes) skim comments. they (sometimes) ask customers. But nobody is systematically turning market language into business assets. 5 niches you could sell this to: 1. Dentists Sources: - Google reviews - Reddit threads - competitor websites - local Facebook groups - patient FAQs Build: "Patient Objection Miner" Output: - top fears - service questions - ad angles - landing page copy - content ideas 2. Gyms Sources: - member reviews - cancellation reasons - competitor offers - local fitness groups Build: "Churn + Offer Insight Report" Output: - why people join - why people quit - what offers pull attention - what testimonials to collect 3. Med spas Sources: - TikTok comments - Google reviews - competitor promos - consult questions Build: "Consult Question + Content Engine" Output: - FAQs - trust objections - offer angles - follow-up scripts 4. Ecommerce brands Sources: - Amazon reviews - competitor reviews - support tickets - ad comments Build: "Customer Voice Mining Skill" Output: - product issues - hooks - objections - comparison angles - new product ideas 5. Agencies Sources: - sales calls - lost-deal notes - client emails - industry posts Build: "Niche Demand Map" Output: - what buyers care about - what they ignore - what language they use - what offer to lead with Charge $1-$3K to build the first research system. Charge $500/mo for monthly updates. This is a high-value system that turns messy market signals into assets the business can use.

• 
  UBU (@pixycup) reported 59 minutes ago

  @AmiriKing Easy fix for Amazon…..those areas can pick their **** up at designated pickup points…not even worth this crap.

• 
  Dutch Gradient (@DGannonTN) reported 59 minutes ago

  @nerd_cookies @antiderivative1 The problem is, the show runners and stars are all-in on SG-U, which was absolute *sh!t* and they were going to base the new show on that dreck. Will Amazon do worse? Maybe, but not by much!

• 
  ZippyTheChicken@GAB 🇺🇸 (@ZippyTheChicken) reported 1 hour ago

  @AmazonHelp I would be willing to leave it at my door but I can not spend 2 hours driving to town with medical issues. I returned the items. I have video of you taking and driving away with them then delivering them again the next night at 7pm Speak here now.. or I consider them abandoned

• 
  Jose Paul Martin (@jpmartin) reported 1 hour ago

  Credit card hacks that actually work in India (from someone saving lakhs on business class & 5-star stays): 1) Best card if you spend under ₹50K/month - SBI PhonePe Select Black. 10% back on utilities & travel, 5% on all online spends. Blended 7-8% return. No other card comes close at this spend level. 2) Spending ₹70-80K/month? HSBC Travel One. ₹5,900 fee waived at ₹8L annual spend. Points transfer 1:1 to Accor hotels. Expect 3-4 free international hotel nights per year. 3) The real game-changer - HDFC Infinia’s voucher hack. Buy Amazon/Big Basket/Croma vouchers on Infinia. Get 12-16% back in points. Even ₹50K/month in daily spending = 80,000 points/year. 4) Those 80,000 points? Transfer 1:1 to Singapore Airlines. Book business class to Japan. A ₹3L ticket for ₹6L of grocery spending you’d do anyway. That’s 50% return on redemption. 5) Japan on a budget without points - ANA’s Hello Blue Sale. ₹40K round trip Delhi-Tokyo. They throw in a free domestic flight (Tokyo to Osaka). Direct flight, 9.5 hours. Comes 3-4 times a year. 6) Biggest hack - book award flights from hub cities, not India. Qatar Q Suite: 160K points from Mumbai. 70K points from Doha (off-peak). Add a ₹15K Indigo positioning flight. Save 90K points on a single ticket. 7) Axis Atlas & Burgundy aren’t dead. Accor transfer is paused but Air India works - Bali for 12K points (normally ₹30-35K). Cards are harder to use, not worse. Expect Accor to return at adjusted ratios. 8) Magnus Burgundy tip - 5% return on first ₹1.5L, then ~14% above that. Best reserved for high-spend months (car down payments, big electronics). Blended 9-10% at ₹3L/month. 9) Amex Platinum Charge (₹78K fee) - worth it in year one (₹60K welcome benefits). After that, value comes from 5x vouchers, 10x Air India, 20x luxury partners. Real perk is Centurion Lounge access for family. 10) The trap nobody talks about - lifestyle inflation. Free flights & hotels shift your mental budget. You “save” ₹60K on flights, then spend ₹1.5L on the trip instead of ₹40K. The math only works if you hold the line. Key rule - only chase deals you’d actually use.

• 
  OldmanC73 (@OldmanC73) reported 1 hour ago

  @AmazonHelp @amazon I already applied for the refund. But like I keep telling any agents the refund doesn't fix the incompetence of those who work at every level in the company nor does it get me the product that I bought. And this continuing problem just shows the lack of concern at Amazon.