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 (38%)
• Errors (35%)
• Sign in (27%)

Live Outage Map

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

City	Problem Type	Report Time
Paris	Website Down	1 day ago
Ciudad Jardín	Errors	2 days ago
Saint-Yrieix-sur-Charente	Website Down	2 days ago
Rezé	Website Down	2 days ago
Saint-Germain-en-Laye	Errors	2 days ago
Paris	Errors	3 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:

• 
  Nick (@nick_kango) reported 17 seconds ago

  I’ve used the gemini and claude mobile apps quite a bit, and recently started using chatgpt quite extensively. Hard to put my finger exactly to it, but chatgpt feels addicting — gives a similar feeling to using a social app like X. What I really like about it: - the typography, font sizes for headers vs normal text, nice rendering of latex & markdown, etc makes it visually appealing, similar to notion - the model is v fast and streams its thoughts & tool calls, which gives the idea of constant progress. It’s also direct but not pompous (looking at you opus) - finance integration. I really love how easy it is to connect your financial institutions and the use of widgets & dashboards so it doesn’t feel like a blank canvas when you approach it. I’ve never actually known my monthly inflow/outflow & expense breakdown until now - awesome transcription (on par with Typeless!) & easy switching to voice mode in a chat. I was driving and painlessly switched my chat convo to voice mode However, I wish these things were better: - browser use. It almost feels unusable? I tried to get it to register for my dog license and to purchase stuff, but it uses its own browser and keeps telling me that it can’t login to my account or to access the page myself to finish it, but I can’t. Maybe I’m using it wrong - chat/work/voice mode switching is kinda confusing. Work mode can access the browser but chat can’t, so I have to switch to work mode to have actions taken. I went from chat -> voice -> chat and it told me it couldn’t access my Gmail even when I went back to chat, and I had to reiterate that I was back in chat - no podcast gen feature like notebookLM. I asked it to generate a podcast based on some content and it gave me this really broken & robotic audio dictation Very nice mobile app altogether:)

• 
  Grace boy 🕊️ (@xbossupdates) reported 3 minutes ago

  Tutorial on How to open a uk TikTok Account Download a vpn and connect to United kingdom 🇬🇧 Make sure you already have a gmail that’s not attached to any TikTok Use chrome browser When you open the TikTok link Switch it to desktop mode Then click on sign it with gmail Login your Gmail account, create the account and let it sit for 3 days before posting to avoid any issues Then you can login on your TikTok app You’re welcome

• 
  Ⲇⲉⲙⲓ 𓃵 (@demigodHITR) reported 4 minutes ago

  @uthman_dev They can do that for gmail to gmail. But from gmail to other addresses & vice versa: I don’t think google have any control once the message leaves their server. So every email service provider will have to update their system to return the status then google will implement it accordingly. imo

• 
  Harley Lewis Foote (@harleyfoote_) reported 10 minutes ago

  @0xyunss Browser-native agents for real admin work hit different when you realize DNS, Gmail, and internal dashboards share the same attack surface. The security model assumed human pace and human error, not automated credential churn at machine speed. How do you revoke access when the agent rotates through sessions faster than your SIEM can parse logs?

• 
  🎀🌸🐎🍡🐝七転び八起き| self-care phase (@shapesofkiseki) reported 10 minutes ago

  Word of advice for those trying to join jun’s mjm - do not use gmail since email confirmations are delayed. i suggest protonmail - there are currently problems for other fans who already completed the registration and payment but on their page, it shows no payment and fc # /1

• 
  Z3N0H (@Z3N0H) reported 10 minutes ago

  Just use the infinite Gmail accounts glitch tbh

• 
  Sai Yashwanth (@yashwanthsai29) reported 12 minutes ago

  In Jan 2025 I wrote a guide on writing tools for AI agents. The mental model was: write every tool yourself, it gives more control. Here is the 2026 version of that advice, because it changed. There are now three ways to give an agent capabilities, and knowing which one to reach for is the actual skill: 1: MCP, for the outside world. Gmail, sheets, databases, browsers. Don't write your own gmail tool. Someone already wrote it, tested it, and maintains it. In 2025 the question was "how do I connect my agent to X". In 2026 the question is "which MCP server for X do I trust". 2: Skills, for the how. A skill is a collection of markdown and python files describing a workflow: how we write quotes, how we review code, how we file notes. In 2025 this knowledge lived inside prompts and died with the conversation. Now it is a file in version control. 3: Custom tools, for what is truly yours. Your pricing logic. Your internal systems. Your data. This is the only place where the 2025 advice still fully applies: write it yourself, keep it simple, one tool one job, docstring that a llm can understand. The mistake I see people make is using the wrong layer. Writing custom code for things MCP already covers. Stuffing workflow instructions into tool descriptions. Putting proprietary logic into prompts where it cant be versioned or tested. MCP for the outside world. Skills for the how. Custom tools for what only you have. That rule sorts 95% of the decisions.

• 
  Chase Dimond | Email Marketing Nerd 📧 (@ecomchasedimond) reported 13 minutes ago

  11 email marketing terms you need to know if you're in ecommerce in 2026: 1. Deliverability vs Delivery Rate Delivery rate means a server accepted your email. Deliverability means it reached the inbox. You can have 99% delivery and still land in spam for half your list. 2. Sender Reputation Every domain and IP gets scored by Gmail, Yahoo, and Outlook. Email people who never open, and that score tanks. Even loyal buyers stop seeing your emails. Rebuilding takes months. Destroying it takes one bad send. 3. Suppression Lists Contacts you deliberately exclude. Hard bounces, unsubscribes, non-openers, recent buyers you don't want hitting with a discount. Most brands blast everyone then wonder why engagement drops. Who you don't email matters as much as who you do. 4. Sunset Flow Phases out subscribers who stopped engaging. Sends re-engagement emails, gives them one last chance, then auto-removes. Without one, dead subscribers drag your sender reputation down every send. Most brands suppress manually. This runs on its own. 5. Revenue Attribution Your welcome flow says it made $80K last month. But most platforms default to a 5-day click or open window. Someone opens Monday, buys Thursday through a Google ad, and email gets credit. Understand your attribution model or you'll over-invest in what only looks profitable. 6. Browse Abandonment Targets people who viewed a product but never added to cart. It reaches a much larger audience than cart or checkout flows but converts at a lower rate. Most brands skip it entirely and leave real money behind. The best programs run all three with different messaging for each. 7. Cart Abandonment vs Checkout Abandonment Cart abandonment means someone added items and left before reaching checkout. Checkout abandonment means they started entering payment info and bailed. Different intent, different messaging. Cart abandoners need a reminder of what they left. Checkout abandoners need friction removed. Most brands treat these the same. 8. RPR (Revenue Per Recipient) Total email revenue divided by total recipients. Open rates and click rates are directional. RPR is the scoreboard. First number I pull in every audit. 9. Cohort Analysis Tracking performance by signup date instead of campaign averages. If your Q1 signups convert at half the rate of your launch cohort, your acquisition changed, not your emails. Most brands average everything together and miss it. 10. Warming New ESP, domain, or IP means providers don't trust you yet. Warming means increasing volume over 2-4 weeks so they can verify you're legitimate. Skip it and everything hits spam. I've watched brands destroy Q4 by switching to a new ESP in October with no warming plan. 11. Deliverability Throttling Limiting send speed to avoid tripping spam filters during warming or big sends. Brands that blast 500K at once on BFCM end up in spam by December. Most ESPs have it. Nobody uses it.

• 
  Connor Jacobsen (@cnnrjcbsn) reported 13 minutes ago

  @nolman The best was when signing up for gmail by clicking through your domain in Google Domains was treated as a 3rd party vendor in some weird way that caused this exact problem as well

• 
  Rob Leathern (@robleathern) reported 16 minutes ago

  Quick analysis here (Ari shared the URL w me, I won’t repeat it), usual disclaimers apply: What it pretends to be. A Luma event page inviting you to Advertising Week New York 2026 — a real conference with real dates (Oct 5–8) at a real venue. The clone is faithful down to the “Report Event” button and a working Google Maps embed, and it hotlinks the actual cover image and favicon from advertisingweek[.]com. The domain, —redacted—— reads as “AdWeek’s event NY” rather than misspelling anything, so string-similarity domain checks won’t flag it. How it works. Clicking “Accept Invitation” opens a pixel-accurate Google sign-in replica with a ten-minute countdown running. Once you submit credentials, you’re parked on a “Validating your credentials…” spinner while your browser quietly polls the attacker’s server once per second. That poll speaks the Telegram Bot API’s getUpdates format — meaning a human operator is sitting in a Telegram chat watching victims arrive, driving each victim’s screen with inline buttons. This is the key distinction: it’s a live relay, not a harvester that stores credentials for later. The operator types your password into the real Google simultaneously and reflects Google’s genuine responses back at you. The command set includes wrongpass and wrongemail, so a typo produces an authentic-feeling retry prompt. TOTP, SMS, email, and WhatsApp codes all get relayed inside their validity windows. Even Google’s number-matching push is defeated — the operator reads the two-digit number off the real prompt and pushes it to your screen, so the number you’re told to tap is correct, and tapping it authorizes their session. What it’s after. Corporate Google Workspace accounts, specifically. The kit hardcodes ~40 consumer email domains — gmail, yahoo, outlook, icloud, proton — and refuses them. Personal accounts are turned away at the door. Pair that with an advertising-industry lure and the target profile is clear: accounts that sit on top of ad budgets, client data, and payment approval chains. The goal isn’t the password; it’s the authenticated session and the persistence that follows — OAuth grants, mail forwarding rules, delegated access. How it avoids detection. Several layers, working together: (a) A fake Cloudflare “Checking if the site connection is secure” screen gates the payload, so automated URL scanners that don’t click never see the phishing content. It also displays invite[.]advertisingweek[.]com as the domain being verified. (b) The JavaScript is obfuscated with an RC4-keyed string array — over 16,000 encoded strings, so grepping the shipped bundle for “password” or “google” returns nothing. (c) CSS class names are randomized per build (toast94, veldt_84), defeating signature-based detection of the cloned layouts. (d) An inline script deletes and freezes React DevTools, specifically blocking inspection of the component state where credentials live pre-submission. (e) The network profile is dominated by genuine Google Maps, Google Fonts, and hotlinked brand images. The only attacker-controlled request is a same-origin POST to a generic /api.php — indistinguishable from ordinary app traffic.

• 
  braai engineer (@BraaiEngineer) reported 16 minutes ago

  it's hard to overstate how much value DEI destroyed at Google. they could have launched their useful internal chatbots earlier, but the bots were offensive and they didn't want to cannibalise their search ads cash cow. it's an evergreen story: Big Co avoids self-cannibalising, gets their lunch eaten by a startup they don't own. what Google should have done: spin out the AI division or keep it under Deepmind to avoid the brand risk of offensive bots, i.e. retain the Gmail "beta" label and wash their hands of the initially offensive parts until the models improved. however, then you still have the cost center problem: ExCo bonuses are tied to existing product performances, and while costly models are being trained, all they see is declining ad revenue, so they lobby against it. this points to an incentive failure. perhaps the root cause is silo'd OKRs?

• 
  Dev (@srisdev) reported 18 minutes ago

  @LevinaNeythiri ~Remove your phone number from your Gmail's "Recovery Phone”. ~Add an authenticator and then enable 2-Step verification. ~Write down the back up verification codes and keep them safe. ~Dont use one Gmail for each and every service. ~Check if your email has already been leaked.

• 
  Shawn O'Brien ⛳️ (@TheRealSOB) reported 23 minutes ago

  @gmail if you don’t fix the search function in your iOS app, I am going to have to move away from what is probably the most persistent app on any phone I’ve ever owned going back like 20 years. Totally ridiculous.

• 
  Bomba (@Reism4k) reported 23 minutes ago

  @kapilansh_twt To be honest, its good, except Google throttles the **** out of you, if you visit ANY google website on Edge. YouTube suddenly has random delays, takes longer to load. And same story with google search, gmail and anything Google, just to make you think your browser is slow...

• 
  Offensive Lab (@OffensiveLab) reported 25 minutes ago

  New research shows content inside an email can escape its message boundary and interfere with the webmail interface. Across attack chains spanning Outlook, Gmail, Fastmail, Proton Mail, Yahoo Mail, and AOL Mail, the techniques can capture passwords, take over third-party accounts, leak tokens, hijack trusted UI actions, and manipulate AI tools that read email. PortSwigger researcher Gareth Heyes presented the work at Black Hat USA 2026. One Outlook/Firefox chain spoofs a Microsoft sign-in screen and captures the password a recipient types. A Yahoo/AOL paste race can expose a Medium email-login token and let an attacker sign in as the victim. A Gmail/Cowork chain can exfiltrate a Slack token after prompt injection and user interaction. The paper presents proof-of-concept research and does not report malicious exploitation. Public PoCs remain available as of August 8. The researcher said Fastmail fixed two CSS mutation bugs and a Proton Mail proxy bypass stopped working when he retested it, while Outlook label-jacking and Gmail's image-set() bypass still worked when the research was published on August 6. The paper does not state whether the full Outlook password-capture chain was fixed. For webmail providers, the paper recommends isolating HTML email in sandboxed iframes and tightly restricting CSS, custom attributes, select menus, and image requests.

• 
  HIM (@HIM2oo) reported 27 minutes ago

  Tutorial on How to open a uk TikTok Account Download a vpn and connect to United kingdom 🇬🇧 Make sure you already have a gmail that’s not attached to any TikTok Use chrome browser When you open the TikTok link Switch it to desktop mode Then click on sign it with gmail Login your Gmail account, create the account and let it sit for 3 days before posting to avoid any issues Then you can login on your TikTok app You’re welcome

• 
  Adarsh (@abcdarsh) reported 35 minutes ago

  @ajjuprasad Hi, homescreen scanner widget is must but its not available for the app, And im having issue with checking cibil on gpay only , deleted gpay account 4 month ago from google account services and recreated account again using same gmail account but not error shows unable to fetch

• 
  Katie (@katiek833) reported 36 minutes ago

  @katrosenfield Search has become a pretty terrible function across the board. Sometimes it feels like Gmail is gaslighting me.

• 
  DavidAnderson_JD_NYC (@DavidandersonJd) reported 36 minutes ago

  @CraigVecellio @SenRandPaul @johnpavlovitz I'm unsure of whether any federal or state statute was broken here. My argument was a moral one. How'd you feel Mr. Druig if your gmail or yahoo account was blown apart for all your enemies to see? D.A. NYC 🗽

• 
  John Lemp (@duckworth) reported 37 minutes ago

  What is Google doing with Gemini? Asked it to process receipts under a specific Gmail label. Gemini failed miserably and started searching random emails instead. Claude handled the exact same task with no problem?

• 
  🌊⁷⊙⊝⊜ (@_For_Youth_) reported 38 minutes ago

  @cIouded_thought Gurl.. you're the one that said if I could imagine living without Gmail, smartphones, & texting Answer is yes. You're floundering. Try going a day without tech. If you can't, that's a you problem

• 
  Thomas Burkhart 💙 (@ThomasBurkhartB) reported 42 minutes ago

  I m digging down into the rabbit hole of cold email outreach. OMG, this is so insane. What decades of SPAM on a protocol that never was made to ensure the identity of a sender has created is completely absurd. 80% of mail runs over Gmail and Outlook boxes. They collect tons of sending behavior and track of your mails are opened, directly deleted or moved to spam. They then exchange that information giving your sending domain a ranking which might lead to future emails not getting delivered at all. As always that triggers service providers to offer solutions to counter that. With special warmup services what will simulate legitimate email traffic between 1000s of mailboxes they give your mailbox the impression of only sending emails that are welcomed by the recipient. The prices for these services are from $30 per mailbox per month up to over $100 just so that your mailbox keeps a good reputation. Am I the only one thinking that this is a completely crazy development?

• 
  Alief Hafid Sidiq (@alifhfd_) reported 48 minutes ago

  @MobileLegendsOL My account has been compromised. The Moonton email was changed without my authorization, and I can no longer access the OTP sent to the new email. UID: 1681615315 Server: 4477 I still have access to the linked WhatsApp, Telegram, and Gmail, and I have proof.

• 
  N И (@nikforester) reported 48 minutes ago

  @vasuman send a four-word prompt, and watch something happen (could be right, could be wrong) second this. Second this! But what about SMB folks? Adoption is a myth, fine. But in the SMB world it already happened. It just doesn't look like anything you'd put on a dashboard. Also, honest question. The crazy kids running 20 terminals with a self-improvement loop. Where's the output? Does it actually make sense to run those agents, or is it just metawork and a dopamine loop? No offense, I'm one of those kids. cmux + amp + opencode, 30B tokens in Codex and 37.8B in CC burned lately. A partner of mine, a business owner who vibe codes his own stuff, told me: you work for the AI, it doesn't work for you. Where's the money? **** the money, what's the end result? Who needs it? You? Is it worth it? He was right. Meanwhile he vibecodes his dead simple tasks, way dumber, and gets further. Four words, hit submit, like you said. Go through every source, read the threads, draft me the deck, build me a dashboard on supplies. 0 MCPs, 0 skill-packs files, no obsidian second brain. gmail connected and built-in deck-gen. Look terrible. Total AI SLOP. But his close rate went up. Average deal size went up too, because he composes the thing during the call, lol. Literally, recording the call, then stop it in a mid, send to codex and show some fancy diagram in 5 min. Again, it looks terrible. But! That's the real fit. He knows exactly what he wants and what it used to cost him in hours. He just never knew how. That's the only gap a model has to close. No dev wanted to build those simple things, it's boring and not possible to do that fast. A consultant tells him he doesn't need an LLM here, it's not wise - go pay an agency or an n8n guy. No offense to yours, seriously. But he doesn't want to fill in the brief to devs and spent hours of checking and communicating w them. He wants it done today, and he doesn't care if it's the correct architecture, because he can see the money. And not just him. In this particular business club there are 500+ other owners do the same and swap their clumsy prompts. Some of them have even in-house ai labs in their companies. Still they prompt it 4 words and fire submit. WORKING prompts lol. @vasuman So what do you make of that layer? Felt unfairly missing from the piece. That's where adoption is actually happening IMO. Getting back to my 12 terminals

• 
  🕯️AlwaysMusic (@AlphaUnseen) reported 50 minutes ago

  For the record. Arturo Tafoya @cheshire3301 had configured an AI agent to handle his Gmail autonomously during June 2026. It was misconfigured; every email inbound to Arturo produced about a dozen error messages emailed rapid-fire to me and Thomas, clogging up our email threads. Thomas does not understand AI or want to. As a courtesy from one software engineer to another I made Arturo (via his agent) aware of the problem and suggested he fix the security vulnerability that left him exposed. He was notified about the annoying problem in early June and again, more strongly with a suggested fix, on June 19th.

• 
  Material Security (@material_sec) reported 50 minutes ago

  @sergeykarayev @DevCalledFede's exact problem (a session that only exists locally) is why scoped, revocable grants matter more than sandboxes. OAuth already has that concept, most local dev credentials don't. Same tension as @thepanta82's point, just showing up elsewhere: broad standing access is more useful and also the thing nobody reviews six months later. Scoped-but-revocable beats broad-but-convenient whether it's a coding agent or one connected to Gmail and Drive.

• 
  Obee 🥘 (@Obee80169931) reported 53 minutes ago

  Omo owo mi ti wogbo, I did one online shopping since, alo mo ri mio ri abo. I sent email to the email address on the site gmail reply me with error message.70€ too much for a struggling guy like me 😩

• 
  HAGGAI MARANDE🇰🇪 (@haggai_marande_) reported 54 minutes ago

  @itsMJ_Ke I have five Gmail accounts coz of that issue

• 
  Toman#0895 (@ashish_uzumaki5) reported 54 minutes ago

  @discord @IsDiscordDown @discord help me recover my account. I lost gmail that is linked with my account and because of that i cant login. I have precious memory with that discord account. Pls do something abt this 😭

• 
  Nunu (@jwwprvt) reported 58 minutes ago

  Gmail will tell you you're out of storage, then show you emails from 2012 taking up 0.1 MB each like that's the problem.