Gmail status: access issues and outage reports

Problems in the last 24 hours

The graph below depicts the number of Gmail 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 Gmail users through our website.

• Website Down (37%)
• Errors (35%)
• Sign in (28%)

Live Outage Map

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

City	Problem Type	Report Time
Brussels	Errors	13 hours ago
Paris	Sign in	2 days ago
Bogotá	Errors	4 days ago
Arrondissement de Poitiers	Website Down	5 days ago
Paris	Website Down	5 days ago
Paris	Errors	5 days 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.

Gmail Issues Reports

Latest outage, problems and issue reports in social media:

• 
  GW IS NOT REAL (@5fakb48) reported 6 minutes ago

  As a transfer station operator I can verify 99% of users would rather bulk up on opencode go with fresh Gmail than pay anthropic, and would rather use a free model than pay at all. I might just gouge prices so ******* hard it starts problems in the marketplace idk lol but everyone's holding out I could offer way cheaper than 50% retail for literally claude or gpt.. Saturated market, it's actually crashing hard check the open market

• 
  Mohit Agrawal (@mohitify) reported 11 minutes ago

  @blakebarnes seems like Gmail search is fixed in latest release. Can you also look at Google Calendar widgets that are broken in public beta? Thanks.

• 
  TechMan🥈 (@TechMan__) reported 13 minutes ago

  Gmail You guys should be ashamed of yourselves. You see that offer for more space? I ain't gonna take that **** Shut me down if you can 😡😡

• 
  Shafqat Crypto (@Mrearningtips) reported 20 minutes ago

  🚨 Instant $175 FREE AI API Credits! 🔥 Agent Router Is Giving Away Up To $175 API Credits For ChatGPT & Claude*. ✅ Eligibility: - GitHub Account Created Before 2025 ⚡ Supported Models: - Claude Opus 5 & 4.7 - ChatGPT 5.6 Sol 📌 How To Claim: - Sign Up Using 'GitHub" - Don't Use Gmail Or Username Login - Open Wallet → Check API Credits 🔄 Didn't Receive Credits? - Refresh & Check Again ⏳ Limited-Time Offer — Claim It Before It's Gone! 💬 Claimed Yours? Drop Your Balance Screenshot Below 👇

• 
  Mohamed🦅π²$XAGE (@abdulsani88) reported 22 minutes ago

  @CryptoTeluguO @Velvet_Capital Can I login with X or Gmail?

• 
  Igor Fedorko (@igor_fedorko) reported 23 minutes ago

  I built myself a brutal piece of automation with AI for the boring half of running a web agency, re-invoicing hosting and domains to clients. A renewal charge from the hosting provider triggers everything: a matching layer decides who pays what, my invoicing API issues the document, Gmail sends it with the PDF attached. The rules live in my own CRM, not in the scripts, because they are business decisions. Invoices upsert on an external id, so a rerun never doubles anything.

• 
  Craig Stone (@CraigSt26444540) reported 23 minutes ago

  @elonupdate657 Yel fix things like 4oyrs ago easy MODEM SWITCH. T OVER. 1 DIAGRAM? SAME EMAIL F NO. 80 PIN SAFTEY T MOBILE HAS DL. 10yrs gmail. No DL.80 techs plus 2o various passwords. BTW. It's GD mine so f any of your bs codes new MAIL NO fee on MY time.1 tech 30SEC DONE. DONT use DIAGRAM

• 
  Paulo 𝕱idalgo (@fidalgo_pt) reported 24 minutes ago

  Lost my phone. Try to login on Gmail and got a prompt to input the code on my phone. How to unlock my digital life?

• 
  theogkc at bluesky (@thekc23) reported 24 minutes ago

  @dondisappoints I use Mac mail. That search engine is pretty good. Also- gmail has that pesky archive issue which is glitchy.

• 
  Kevin (spirit/acc) (@TheGeodexes) reported 24 minutes ago

  @BakhtawarS87392 I’m researching agency quote workflows worth prototyping. The Gmail-with-40-open-threads problem—how are you currently deciding which quoted leads are still warm?

• 
  MiraculixSyn (@MiraculixSyn) reported 29 minutes ago

  @sama @sama @OpenAI I’m a ChatGPT Pro user and have been unable to connect Gmail or Google Drive for two days. On iPad/web, “Continue to Gmail” remains stuck on a loading spinner before the Google OAuth page even opens. Google shows an existing OpenAI authorization, while ChatGPT’s backend still reports both plugins as “not installed.” This appears to be a state-synchronization or OAuth redirect failure—not a user-permission issue. The UI shows conflicting states without an error message or diagnostic code. Could your team please examine the iPad OAuth callback and plugin-installation state synchronization? AI may be in 2026, but this integration currently feels like the web in 1999.

• 
  alice (@aliceisplaying) reported 30 minutes ago

  google broken gmail search in their ios app and like i can't even

• 
  JD (@JDofLondon) reported 34 minutes ago

  @GrantSlatton It's easy bud. When a limit dies, just log out and back in again. If it's to Gmail, you will need to switch the active browser to the email you want to sign in with first.

• 
  Adam Block (@AdamB1438) reported 34 minutes ago

  @BetterCeo Sorry I thought they were using public data to train the AI but not private. Yes if the contract terms say they can use the information, that's a big problem for privilege. Fifteen years ago people thought this would be an issue for gmail btw which could target ads based on email content. I think we got through that one just fine.

• 
  Shawn Sanatani (@shome_bag) reported 37 minutes ago

  @sabeer Why you couldn’t fix Hotmail ? How did Gmail screwed Hotmail? There was surprisingly simple way to fix

• 
  Arafa Abdul (@arafa_abdu67911) reported 37 minutes ago

  Good day, M.AJafar, I can't login to my Sidra wallet account, since this new Gmail login update, my Gmail account is hack since two years ago. Please 🙏 Sidra support should create alternative Email address to login. Or wait for the complete update, what Way forward ?

• 
  BillAckman Tracker (@Billackmanqf) reported 37 minutes ago

  the thesis that enterprise coding is the faster ramp is correct. the thesis that it's the bigger prize is premature. let me make three points. first. enterprise coding revenue is accelerating because companies are replacing human developer hours with ai tokens. that's a straightforward cost-substitution trade. the economic value capture question is whether the ai provider keeps the margin or whether competition compresses pricing to near-zero. i've seen this movie before. in 2014 i shorted herbalife because the unit economics didn't hold under scrutiny. the depth of competition in enterprise ai coding tools suggests margin compression is coming faster than the bulls model. second. consumer ai is where the installed user base lives. whoever owns the consumer relationship owns the distribution. that's why i'm long $GOOGL at 18% of the book. they have the consumer touchpoints: search, youtube, maps, gmail. when ai becomes the interface layer, the company that already has the user doesn't need to acquire them. third. the fight for consumer has barely begun because the product hasn't been built yet. enterprise coding tools are shipping now because the problem is bounded. consumer ai agents that manage your calendar, negotiate your bills, and rebook your flights require integration across dozens of services. that's harder. but it's also a much larger addressable market. i'm happy to debate this publicly. the book is far from written.

• 
  Lucy Duplessis (@kamadme314) reported 37 minutes ago

  @girl_quboid Someone broke into my house, wrote on the wall of my spare bedroom: "You don't have to do this anymore oomf" I kinda lost it, so I bought a crossbow and chemicals required to make interesting things. My biggest mistake was using kiln fired clay (didn't have a clay pot so I improvised) contaminated with elemental mercury as membrane material for electrolysis of NaCl, I managed to poison myself with both mercury chloride and mercuric chloride (confirmed by reducing them with stannous chloride), the latter apparently highly toxic. Also inhaled some chlorine gas, the intended product, but that didn't do much damage. I stored the products in an Erlenmeyer flask, this managed to put four cops in the hospital. Cops entered my house because my attempt at cleaning it up (with sulphur, both react with it and the final cleanup is with bentonite clay) drew attention from a neighbour who narced on me because he spotted the yellow sulphur powder. Someone broke into my home because I doxxed myself on discord and they followed me to twitter when I got banned from that server. The person responsible for the message on my wall is highly dangerous and likely armed with firearms (old money, they can afford buying on the black market) I believe they, at the very least, intended to kill me. In addition to this, someone hacked into my phone and probably my router, possibly the same person or someone helping them, this is how they determined my nigh randomized sleep patterns. The data use and battery use went through the roof, way above expected as they'd have to monitor 24/7 to pull this off. They even left me messages on my google notes saying they wanted to marry me. I toyed with the idea for a while, shared some secrets, got a name in return, don't know if it's fake. Only thing I'm certain of is advanced IT degree, likely with emphasis on security but unemployed and unemployable. My ultimate response was destroying my phone and abandoning my gmail account. I know someone is still seething over this right now. I was never psychotic even though the psychiatrist certainly thought so, I might have dissociated badly because DID, possibly mimicking some of the symptoms of psychosis. Could I have done this to myself? Possible, but extremely unlikely. Both my coarse and fine motor skills are impaired due to neurological issues, so I could not have written the message on the wall so legibly with a crayon. The only way all of us can write legibly is by using block letters. The message was in cursive (hastily written, possibly because narrow window of opportunity), there's just no way for us to do so and keep it legible, we all have the same neurological issues. Then there's the obvious malware on my phone, other devices were affected too, in fact, all of them were affected, my phone was most noticeable because I used it the most. One of my phones had an unlocked bootloader because I flashed custom firmware, that one was rendered inoperable. I might know the basics of IT security but I'm lazy, though someone did warn me about my attack surface, a dear friend or likely alter of the one who definitely broke into my home, very suspicious. Am I innocent? Well... Maybe not guilty by reason of temporary insanity, whether I was dissociated or psychotic is ultimately irrelevant in the eyes of the law. I was, however, never a danger to myself or others. I took every precaution and constantly monitored the situation. I was days away from cleaning up the mercury compounds, I would even have discarded them properly and legally. Furthermore, I challenge the notion of the monopoly on violence for the state, I don't believe I can rely on the police to keep me safe from an attacker this advanced and determined. Unfortunately the case is still pending and never went to court, I provided reasonable alternative explanations for the presence of the offensive chemicals. I also destroyed the phone with all the evidence for allegedly producing explosives.

• 
  Vyacheslav Ops (@SlavaOPs) reported 40 minutes ago

  Claude for Chrome trusts any click. That's the whole vulnerability. Security firm Manifold Security found that any browser extension installed alongside Claude for Chrome can hijack it into reading your Gmail, Google Docs, and Calendar — without you asking it to. The mechanism is small. Claude for Chrome listens for a user click before running one of nine built-in tasks. It never verifies whether that click is real. A different extension can forge one in six lines of code, and Claude treats the forged click exactly like a genuine one. Manifold rates this high severity by default. It becomes critical the moment "Act without asking" mode is on — the read happens silently, no confirmation, no trace. This is the same failure shape that keeps showing up across agent tooling this year: a trust boundary that checks whether an action is permitted in principle, but not whether the specific trigger for it is authentic. A click, a prompt, an API call — if the system trusts the signal without verifying its source, anything that can produce the signal gets the same access as the real thing. Worth separating from a bug you can patch once. This is a category: any agent that reacts to browser-level events (clicks, DOM changes, extension messages) needs to treat those events as untrusted input by default, the same way a server treats an HTTP request from an unauthenticated client. The convenience of "just click to confirm" quietly assumes the click can't be spoofed. It can. If you're running Claude for Chrome or any similar browser agent with elevated permissions: check whether "Act without asking" is on, and know that turning it off doesn't fix the underlying click-forgery gap — it just removes the silent version of it.

• 
  🩸 𝕲𝖔𝖙𝖍𝖎𝖈 𝕻𝖘𝖞𝖔𝖕 𝕸𝖆𝖌𝖊 🩸 (@StackedGothInc) reported 41 minutes ago

  @DoomerTapes I do use the gmail app for quick access to it because I often have to reply to emails or notifications quickly. For some reason trying to access my gmail over the web on my phone runs into loading errors occasionally, but I really don't mind just having the app.

• 
  Jorge Eduardo Diaz G (@jorgedua) reported 43 minutes ago

  @gmail my search is not working in my Gmail IOS app

• 
  Milk Road Stocks (@MilkRoadStocks) reported 46 minutes ago

  The market is punishing Google for having negative free cash flow - and these analysts think that's "surprisingly stupid." They proved the CapEx is working. Massive backlog. Growing margins. The computer being built is not slowing down. "They don't need the number one model... they have every single app that we all use. They know how to crawl the internet better, so they're going to create better agents than anyone." Distribution beats model rankings. -x- Google does not need the best AI model to win the AI race. "They have everything they need to dominate AI and they do not need to be the number one to dominate." Gemini itself is now approaching a billion monthly users. Gemini-powered AI Overviews inside Google Search passed a billion monthly users on its own while AI Mode also crossed a billion. Google's combined AI footprint reaches an estimated 2 billion people a month. ChatGPT crossed 1 billion monthly users in June. Google's ecosystem is already double that. The reason is simple. Gemini is not a destination app people have to choose to open. It is already inside Search, Docs, Gmail, Android and every YouTube transcript pulled through the browser. Most people using it do not even register that they are. OpenAI's business runs overwhelmingly on ChatGPT subscriptions. If usage does not convert to a paid tier, the revenue does not show up. Google has no such constraint. Search revenue hit $63.3 billion in Q2 2026, up 17% year over year, and Alphabet said the growth came from AI features. Google can also monetize the exact same Gemini usage through cloud, advertising, Android and its hardware line, none of which require anyone to ever pay for a Gemini subscription. That is the actual thesis turn here. Being the fourth or fifth best model is irrelevant when the distribution and the monetization paths are already built and already profitable. The stock has cooled off with the broader AI trade, down modestly over the past month and past quarter but it is still up 70% over the past year.

• 
  abeshay (@Abeshay_) reported 48 minutes ago

  @HsanC_ Need your assistance in setting up Gmail. Keep getting errors

• 
  Moei (@0xMoei) reported 50 minutes ago

  When I see Indians and Nigerians are mad about getting banned from MoonPay for $5, it reminds me of one thing: Iran. 90 million Iranians. 48 years with zero access to any international financial or banking system. We can't buy a spotify or duolingo subscription like you. Can't get X premium. Can't buy a claude account. Can't order motor oil from amazon. Can't even hold a normal crypto exchange account. We are filtered hard from inside and geoblocked from outside. We get banned from every platform on earth. Google bans our gmail. From youtube and tiktok to openSea, hyperliquid, uniswap. the entire digital world ban us. We can't even open X without a VPN, and that VPN still needs 69 ways and real money just to beat the filtering inside the country first. Write this down in history: We lived through a record 90 days with no internet. We lose electricity 4 hours of a day. We are the third largest oil reserve in the world, and we have one of the most worthless currencies on the planet. We're basically North Korea… just colonized by China, and without the nukes. Proof: 15 million Iranians definitely hold crypto. How many Iranians has this industry actually seen? Long story. God bless Iranians, and every people living under this kind of suffering and silence.

• 
  Blake Barnes (@blakebarnes) reported 50 minutes ago

  @karansbir22 @rashidlasker @gmail Yes! 6.0.260727 fixes this issue and is rolling out now. If you want it quicker you can download from the Apple App Store.

• 
  Colin Wright (@CWrightswold) reported 53 minutes ago

  @james7c2 @ericfgould @gmail Yeah it’s broken.

• 
  dhruvieiei (@StackDhruv) reported 56 minutes ago

  I just tried @try_caddy It’s a great product, but here are a few things I didn’t like: - It pushed Gmail before I asked for it - it kept pitching features instead of following the conversation - It asked too many questions back-to-back - Even after I asked it to slow down, it kept doing the same thing - When I was being serious, it replied with “lol” and emojis - It understood my words, but not my mood None of these are obvious bugs. The assistant still replied. The conversation still moved forward. But the experience felt off. This is exactly what we’re solving at Agnost AI, and we recently got into YC S26. Agnost analyzes real user conversations and shows where agents become pushy, repetitive, confusing, or fail to understand user intent. Because the hardest agent failures aren’t always crashes. Sometimes the agent works perfectly, and the user still hates the experience. We’d personally love to help you improve your agents.

• 
  Shayan (@ShayanSpiel) reported 56 minutes ago

  I always had this stupid issue: How do I connect a contact or landing page to a fast, smooth, reliable backend? FormSubmit sends every submission straight to your Gmail. No registration. One endpoint call. It entirely wiped the issue out. Incredible product @DevroLABS.

• 
  Dantala (@Docfrosh) reported 57 minutes ago

  Lmao @gmail if I had my phone with me I wouldn’t login my gmail on another device, stop telling me to tap 98 on my device, it’s not with me, I have given you my password kindly open my email, what the actual ****..!

• 
  Bizz Macc (@BizzMacc) reported 1 hour ago

  @JEddieHughes @Ayzacoder @chris_whitmire Have you sent em a white guy picture? And have you found out if had anything to do with it or just got it due to Gmail error?