Gmail 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:

• 
  Dylan Merigaud (@dylan_merigaud) reported 2 minutes ago

  @zeeg Hit the literal version on Gmail: search results render inside a display:none container, every row a zero-size box, any click on it correctly fails. Fix wasn't a smarter click, it was reading the thread id off the row and going to the permalink directly.

• 
  Amber⁸ ⋆.˚ Gargoyle Candle + Nest & That (@ALD1passenger) reported 3 minutes ago

  if you don't have time but do have a little money, sign in w/ every gmail account you have and buy the $1.99 gem pack and you'll have more than enough for 10 votes

• 
  W. W. S. Hsieh - Sometimes Satirical! (@whsieh) reported 3 minutes ago

  @jwithing UT’s own webpage on Zotero says this so it doesn’t seem to be a configuration error. Also, since UT doesn’t use Gmail, I’m not wedded to using Chrome on my work laptop, anyhow.

• 
  آلیس در سرزمین حقایق (@khargiush) reported 3 minutes ago

  @TeamYouTube This is a misunderstanding! My GOOGLE ACCOUNT/GMAIL was not suspended for violations. I am simply locked out due to identity verification at login. I need a human agent to review my ID before my Google account is mistakenly deleted. Please help! @TeamYouTube @Google

• 
  Divine Ndubuisi (@DivineN87555) reported 7 minutes ago

  @jamb_hq @prosperarinze66 Please i lost my Gmail for jamb login and what can I do in order to print out my admission letter

• 
  Logan (@limeunfiltered) reported 13 minutes ago

  @GreggBurke Me and @truptinatu are having the same issue, been at it for an hour already. @bot only seems to allow using the browser to work with gmail which won't work well for my use cases.

• 
  Wayne (@im_just_ethan) reported 13 minutes ago

  Was having a “woe is me” driving then a scammer almost got me for my Gmail with my whole life on it, once I shut him down I remembered who ******** I am and that my problems are manageable with my circle and my main man JC. Little adversity to remind me I’m him. Thanks scammer

• 
  Mat Velloso (@matvelloso) reported 14 minutes ago

  Hey @madebygoogle, found the root cause: Your August update broke the magnification shortcut: With that enabled the scroll gesture becomes flaky, misfires and fires backwards. Disabling it makes it go away. I can repro that in 3 different devices. You can repro it in most apps if you try to scroll fast enough: Gmail and Instagram in particular. Also: Please fix your support. That was a massive waste of time. Thanks.

• 
  Rick (@pickl0rick) reported 18 minutes ago

  @uwukko Do something about the cookies please I can't login gmail everytime I open the browser

• 
  Rowan G. Lóchrann (@RowanLochrann) reported 20 minutes ago

  @OpenAIDevs : you have a state-machine problem masquerading as an AI product. You built Chat and Work with nearly identical conversational affordances while giving them radically different authority semantics. In one, natural language primarily produces proposed output. In the other, the same natural language can trigger state-changing actions in external systems. Then you expect the user to carry that distinction in their head. That is catastrophically backwards. Execution authority is application state. Application state must be visible. “Draft this,” “change this,” “fix that,” and “try again” are ordinary conversational instructions. They are NOT durable authorization to mutate Gmail, Drive, websites, or anything else with external consequences. An agent capable of external action needs an explicit state machine: ANALYZE → DRAFT → REVIEW → AUTHORIZE → EXECUTE → CONFIRM A correction after EXECUTE must return to DRAFT. It must never inherit execution authority from the previous action. Otherwise an agent can respond to an execution error by autonomously creating another execution error. And moving a conversation into an agentic environment should not silently change the meaning of the user’s verbs. A handoff of context is not a handoff of authority. This is not solved by a smarter model, longer reasoning, better prompting, or—dear God—a Pro upsell for “improved accuracy for tool workflows.” It is solved by competent capability design: least privilege, explicit authorization, visible state, scoped consent, and authorization invalidation after every external action. You are building agents. Start designing the interface like those agents can actually do things.

• 
  A.I.Goni (@elgoney1) reported 22 minutes ago

  @AirtelNigeria better fix up. Why cant I access my email (Gmail) with your network 😡😡😡😡

• 
  kushagra chitkara (@kushagrchitkar) reported 22 minutes ago

  Instinct raised $350M this week. It also spent the week in a privacy backlash: screen access, data retention, actions users didn't confirm. Both things are true, and both matter. The demand signal is undeniable: people want an agent that acts first. The backlash signal is just as clear: nobody wants an agent that watches everything to do it. This is the exact design problem we've been living inside at Almanac. Our answer: the agent doesn't watch your screen. It builds context from where work already lives Gmail, Slack, Notion and acts on real triggers: a meeting on the calendar, a renewal date, an email that arrived. When it texts you, the draft is already done. Not because it was spying. Because it already knew. Proactive without surveillance isn't a compliance checkbox. It's the whole architecture. The market just proved the demand at $2.5B. The trust layer is still up for grabs. That's what we're building.

• 
  Mustafa Ekinci (@ekinciio) reported 25 minutes ago

  SaaS is not dead. AI is separating durable systems of record from replaceable workflow software. That was the most useful takeaway from the latest @theallinpod episode, where the hosts discussed Salesforce’s earnings and why the “SaaSpocalypse” narrative may have gone too far. A few months ago, the dominant argument was simple: AI agents can write software, so companies will rebuild their SaaS tools internally and stop paying vendors. The @theallinpod hosts argue that this misses the difficult part of enterprise software. Building the first version of a CRM with AI might take a weekend. Making it secure, reliable and scalable is a different problem. You still need permissions, compliance, data protection, integrations, auditability and support for thousands of edge cases. Friedberg shared a practical example from his own company. His team built an internal CRM using AI coding tools. The prototype worked, but every new requirement created another layer of work: account security, access controls, protected data storage and additional features. They eventually moved to Salesforce. The company’s engineering time was more valuable when spent on software unique to its business, such as plant breeding and laboratory workflows, rather than rebuilding a generic CRM that already existed. This creates an important distinction: Horizontal SaaS products such as Salesforce, Workday, Oracle, SAP, Slack, Gmail and Excel are difficult to replace because they hold canonical company data and support workflows across many industries. Vertical SaaS products are more exposed when their main value is a narrow workflow that an AI agent can recreate or customize for one company. The strongest SaaS companies will survive because they are systems of record, not because users love their interfaces. In fact, the interface may become less important. The hosts described a future where users work primarily through Claude or another general-purpose AI agent. The agent reads information from Salesforce, performs actions and writes the results back. Salesforce remains the trusted data layer even if Claude owns the user interaction. That explains why Salesforce’s Anthropic integration matters. Benioff appears willing to let Claude become the front end because Salesforce can remain the source of truth behind it. The new enterprise stack could look like this: • A trusted database and system of record • Applications and business workflows • AI agents that use those workflows • A general AI interface controlling the agents This also changes what SaaS founders should build. A polished dashboard is no longer enough. SaaS products need excellent APIs, command-line interfaces, permission systems and agent interfaces. External agents should be able to read data, execute approved actions and write results back safely. Companies that insist users interact only through their proprietary interface may lose relevance. Companies that make their data and workflows useful to outside agents can become more deeply embedded. My takeaway is that AI will not eliminate SaaS as a category. It will expose which products own durable data and which products are only workflow wrappers. The likely winners: • Trusted systems of record • Products with strong retention and low churn • Platforms with years of operational reliability • Companies with excellent APIs and agent access • Tools that help customers use their proprietary data safely The most vulnerable: • Thin vertical SaaS wrappers • Products built around easily replicated workflows • Vendors with weak data ownership • Companies that resist external AI agents • Tools whose main advantage is a user interface The question for every SaaS founder is becoming clearer: If the interface disappears and an AI agent operates the product on behalf of the user, what durable value remains underneath?

• 
  Joe Backchannels (@joebackchannels) reported 27 minutes ago

  High email bounce rates are not a messaging problem. They are a data problem. Between 23% and 30% of B2B email addresses go invalid every year, from job changes, restructuring, and company churn. That means a list you built 12 months ago is already significantly broken before you send a single sequence. Here is what happens next. You send to unverified contacts. Bounce rates spike. You cross the 2-3% threshold that Gmail and Outlook use to flag senders. Once flagged, even your good emails start landing in spam. You have now penalized your entire sending domain because of the bad records at the top of the funnel. 5-10% of B2B email addresses in any given database are invalid right now. Not eventually. Now. That is not a rounding error, it is a structural tax on every campaign you run. The industry's answer is usually list cleaning tools and re-verification workflows. But the cleaner fix is to start with verified contacts in the first place, so you are not playing catch-up with your own list hygiene every quarter. Precision at the source beats cleanup at the end. Every time.

• 
  Dennis Goedegebuure (@TheNextCorner) reported 28 minutes ago

  is gmail down?