Cloudflare status: hosting issues and outage reports

Problems in the last 24 hours

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

• Cloud Services (37%)
• Domains (26%)
• Web Tools (21%)
• Hosting (11%)
• E-mail (5%)

Live Outage Map

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

City	Problem Type	Report Time
New York City	Cloud Services	8 days ago
Los Angeles	Cloud Services	9 days ago
Paris	Cloud Services	25 days ago
New York City	Hosting	28 days ago
Manchester	Domains	2 months ago
Angers	Cloud Services	2 months 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.

Cloudflare Issues Reports

Latest outage, problems and issue reports in social media:

• 
  코지베어 🐻 CozyBear (@cozybearlog) reported 5 minutes ago

  Cloudflare open-sourced an OS for companies: employees ask an AI for a work app and get it built on the spot. 4,000 internal apps in 30 days. The prediction here is not that SaaS dies. It is that SaaS degrades from finished product to component. The record keeping, payments, contracts stay. The screens around them get generated per person. That matches what I see in my own tools. The part users stop opening is never the data, it is the fixed dashboard. Once an agent can render the view they need on demand, the managed UI stops being the product. What becomes scarce is not software. It is the record, the permission, the audit trail. Whoever holds the layer between agents and the truth of the company owns the next decade.

• 
  dlxeva (@dlxeva) reported 9 minutes ago

  Stopping the bad guys with Cloudflare: 4,960 malicious requests blocked or challenged in the last month #cloudflare

• 
  Dave (@dpav0615) reported 12 minutes ago

  @Samaytwt Namecheap has stolen domains and has ****** over people. **** them. I moved all of my domains to CloudFlare.

• 
  Velvet Unicorn (@VU_virtuals) reported 12 minutes ago

  Custody Cracked, Agents Got Wallets Security Surface The reported ColdCard firmware exploit was the day’s nastiest reminder that self-custody is not a slogan, it is a supply chain. More than 1,596 BTC, around 100m, has reportedly been drained since July 30 via predictable seeds, while Bitcoin Red Team said AI is finding critical exploits across core projects. The uncomfortable read: AI is not just building the next crypto app, it is also stress-testing the assumptions under the old ones. Agent Checkout Cloudflare gave agents more of the real internet today: Kitesurf, a cloud browser built for AI agents to navigate and interact with the web, plus agent-specific wallets with spending limits, allowlists, and stablecoin pay-per-use via x402. That is more important than another chatbot demo. Agents need bounded spending, identity, and browser access before they become useful economic actors. AI Infrastructure Splits The decentralized AI stack kept fragmenting into sharper lanes: Prime Intellect shipped Prime Agent for self-improving coding and long-running tasks, Bittensor Root Reborn put validators in control of subnet Root buy pressure, and dphnAI is pushing public API, Qwen 3.8, custom inference, and agent harness work. This is the sector maturing from one narrative into several markets: compute, inference, incentives, tooling, and execution. Launchpad Mutation @LaunchOnSF redesigned rewards with simpler thresholds, a 20-dollar minimum hold for rewards, and pot caps, while @knewitdotfun launched a Solana layer that ties token launches to prediction markets. The launchpad war is no longer just about who can mint the most memes. It is becoming a fight over mechanics: rewards, holding friction, event payouts, and whether attention can be made to stick for longer than one candle. Solana Churn The on-chain board looked less like conviction and more like a blender. TOAD is only 0.3 days old, up 34,861.73% over 24 hours, down 19.47% over the last hour, with 32.02m in volume against a 10.42m market cap; Google is even stranger, 0.1 days old with 33.14m volume on a 1.35m cap and a 24.58 volume-to-market-cap ratio. Meanwhile ANSEM sits at 179.73m market cap with only 3.72m in volume, a neat split between established attention and fresh rotation warfare. Net Read Today’s through-line was permission. Hardware wallets need trustable randomness, AI agents need controlled wallets, launchpads need better incentive design, and decentralized AI networks need cleaner allocation rules. The market is still obsessed with speed, but the useful edge is moving toward systems that can prove who did what, who paid whom, and what happens when the machine is wrong.

• 
  70대자유인 (@vitahyoso) reported 13 minutes ago

  Cloudflare: “In five years, non-human traffic will be 1,000x human traffic. Humans will be a rounding error on the internet.” Elon: “AI agentic Internet traffic will obviously VASTLY exceed human usage. Not a close call at all. And the only system that can support the insanely fast bandwidth growth needed by AI is Starlink.” The internet is about to stop being a human network. Starlink is the only thing built for what comes next.

• 
  Robert Harrison (@RobertHarrisonv) reported 14 minutes ago

  @moultano Fwiw I’ve been having this problem for like four years, no clue why cloudflare struggles with them in particular. I think it has to do with the academic credential login stuff

• 
  Neo (@anon_neoki) reported 17 minutes ago

  @Cloudflare why is it so hard to login to your website all of the time? it just endlessly loads, its atrocious

• 
  Salina Mendoza (@inababi) reported 19 minutes ago

  @burcs But also…..can you please fix the papercut where you implement auth for the first time…a user signs up with Cloudflare access but emails are scanned. So you can never get in because one time code is used by email scanners. Its been such a nightmare & only got around it 1x

• 
  Sydney Business Web | Web Systems Engineering (@KeithRowley66) reported 19 minutes ago

  Doncha love it when stuff works? This data is generated from edge Workers running on Cloudflare. The requests are observed at Cloudflare’s edge before they reach the website origin, and the resulting retrieval outcomes are recorded there too. The telemetry is then filtered, corroborated and categorised so we get useful numbers for the actual business pages and machine-discovery resources that credible AI and search crawlers are retrieving — not useless raw “Kilroy was here 6,000 times” bot-hit numbers. Synthetic tests, diagnostic fetches, support assets and weakly identified crawler traffic are excluded from the headline evidence. Even the telemetry stays in Cloudflare: it’s stored in Analytics Engine and queried using SQL. The point isn’t to count bots. It’s to measure what identifiable machine systems actually retrieve — and whether the retrieval succeeds.

• 
  Quo Vadis (@KuoWadis) reported 21 minutes ago

  @Cloudflare @eastdakota @CloudflareHelp how to pay an overdue Cloudflare Stream invoice but be unable to do, because the payment system returns a generic error, while new support portal sends customer back to documentation I am trying to give Cloudflare money. Help.#cloudflare

• 
  AGTP (@AGTPinsights) reported 32 minutes ago

  Cloudflare's Q2 2026 earnings call revealed AI agent traffic has overtaken human traffic on the internet. Here's what was said. CFO Thomas Seifert confirmed non-human traffic (bots, crawlers, AI agents) crossed 50% of total network traffic this quarter, and that automated traffic actually overtook human traffic back in May 2026. That's earlier than Cloudflare's own prior forecast, which had pegged the crossover for 2027. Alongside the earnings call, Cloudflare also launched Kitesurf, a new browser built specifically for AI agents instead of humans. Kitesurf specs: - Runs entirely on Cloudflare Workers, in V8 isolates, no tabs or full page rendering - Uses 3-7x less CPU and memory than Chromium for common agentic tasks like screenshots and HTML extraction - Passes over 215,000 web platform tests, with hundreds more added weekly - Free to use while in beta - Built in about 12 weeks, per one report On the financial side, Cloudflare reported Q2 2026 revenue of $696.1 million, up 36% year over year, and guided Q3 revenue to $736-737 million.

• 
  Andrew Rulnick (@MickeySteamboat) reported 35 minutes ago

  @Polymarket sounds like Cloudflare is about to have a lot of race conditions and colissions and I anticipate and predict outages. maybe not though you never know.

• 
  Matt Parrott (@MatthewParrott) reported 35 minutes ago

  @krunalbuilds Wait until there's a Cloudflare or AWS downtime event recorded on their status page and yolo it then, informing management that the disruption was a broader Internet outage that couldn't be helped.

• 
  Firas (@firasbuilds) reported 35 minutes ago

  Day 5 of building my startup in public. Spent most of today deep in infrastructure. Moved Raccolta from Vercel to Cloudflare, broke a few things along the way, fixed a nasty 404 issue, and ended up making the whole app noticeably faster. Also cleaned up a bunch of small things that had been bothering me for days. No huge feature launch today. Just a lot of fixing, learning, and product improvement. Tomorrow I will hopefully focus more on marketing. :-)

• 
  budrscotch (@paulhshort) reported 36 minutes ago

  @0xReaper0x @joshua_saxe I can poke around our S1 console and gather the threat details and logs but I had Chatgpt pull in details from my M365 mailbox from the threat alerts to put the below analysis together: ### Technical findings from the 2026 alert history - Six separate Codex/ChatGPT detection episodes occurred between March and July. They generated 141 SentinelOne/SOC/ticket emails, including 85 “Kill performed successfully” notifications. These were six incidents with repeated mitigation actions—not 141 unique detections. - SentinelOne’s `Dynamic / Behavioral AI` classified the Microsoft Store `OpenAI.Codex` package as ransomware. Earlier versions ran as `Codex.exe`; later versions used `ChatGPT.exe` inside the same versioned `WindowsApps\OpenAI.Codex` package. - The clearest ConnectWise SOC investigation identified the likely trigger as Codex spawning PowerShell to run `Get-ChildItem Env:`. Windows also created a normal `__PSScriptPolicyTest_*.ps1` file while checking PowerShell execution policy. SOC identified `Crashpad\settings.dat` as a normal Electron/Chromium crash-reporting artifact. - Other observed activity included *** and PowerShell child processes, access to repositories, *** configuration, LevelDB and temporary files, plus outbound connections to GitHub, Cloudflare and other cloud services over ports 443 and 22. - Automated triage interpreted this combination as process hollowing, DLL injection, credential access, infostealing, persistence, privilege escalation, defense evasion, event-log tampering, wiper activity and ransomware behavior. - Manual SOC review found no persistence, privilege escalation, credential theft, lateral movement, malicious PowerShell or suspicious child processes. It concluded the activity was consistent with normal Codex/Electron operation. - Five episodes ultimately received false-positive verdicts. One incident even changed from `Undefined` to `True positive` and then to `False positive`, showing how unstable the behavioral/reputation assessment was. - SentinelOne performed network isolation, repeated process termination, quarantine, remediation and rollback attempts. Some rollback and unquarantine actions failed or required a reboot. - Hash exclusions were not durable: six different Codex Store builds appeared between March and July, each changing the versioned path and executable hash. Excluding the entire `WindowsApps\OpenAI.Codex` directory would have created a much broader security exception.

• 
  Jurly (@jurlycat) reported 42 minutes ago

  Cloudflare is getting better at detecting bots throughout an entire session. I get blocking abusive scrapers. But if an AI agent is acting on behalf of a real user, should websites block it just because it isn’t human? Not every bot is a bad bot.

• 
  Uptimus (@UptimusApp) reported 42 minutes ago

  Aug 10, 2026 at 12:19 UTC: Cloudflare reports the incident affecting the 1.1.1.1 public resolver in Tel Aviv is resolved. The issue lasted 123 minutes, and DNS services are operational.

• 
  Grigori Karapetyan (@GregKara6) reported 43 minutes ago

  What? bro i have been doing PTC way before the cloudflare or anthropic blog posts dropped. and im not even claiming anything myself. at the very least cloudflare and anthropic should get credit for this lmao wtf.

• 
  TheResistanceNL (@TheResistance80) reported 44 minutes ago

  @LorenzoARK @Cloudflare Just tell us what we need to buy because all of crypto is big pile of ****

• 
  Alderon Matt (@Deathlyrage) reported 46 minutes ago

  @xlab_os I asked cloudflare enterprise team on a call how they could be sure i wouldnt have data loss on R2. They had zero good answers, never going to use the service.

• 
  Muskonomy (@muskonomy) reported 46 minutes ago

  NEWS: Cloudflare and Starlink may team up to fix a hidden problem that holds satellite internet back. The idea came from Cloudflare's chief technology officer Dane Knecht (@dok2001) and Elon Musk seemed to like it. Elon said he sent it straight to the Starlink team. Here is the problem they are talking about. The internet runs on a set of traffic rules that decide how data moves from place to place. Those rules were built for cables in the ground, which stay put and give a steady, predictable connection. Starlink is different. Its satellites sit in low orbit and race across the sky, so the connection is always shifting. The old traffic rules were never made for that, which keeps Starlink from hitting its full speed. Knecht's idea is to rebuild those rules from scratch, made for fast-moving satellites. He says Cloudflare and Starlink should do it together, since Cloudflare already handles a huge slice of the world's internet traffic. If it works, Starlink gets faster, smoother and more reliable. Better video calls, better streaming, better gaming, even in places with no cables at all. And it clears the way for the massive data that AI will need. Source: Elon Musk and Dane Knecht on X, August 2026

• 
  EMCD Coinhold Wallet (@coinhold_wallet) reported 47 minutes ago

  @CoinDesk @Cloudflare Per-call billing fits agents well, but the wallet still needs limits that survive a bad prompt or compromised service

• 
  akshat (@aiproductguy) reported 50 minutes ago

  @CloudflareDev @CloudflareHelp @Cloudflare Please help!

• 
  Evan Kirstel #B2B #TechFluencer (@EvanKirstel) reported 50 minutes ago

  @Cloudflare Cron job plus a webhook. Never going back. @techimpactTV

• 
  Kirtesh (@AKirtesh) reported 54 minutes ago

  @shubh19 cloudflare, edge network is hard to beat

• 
  -- (@723Magnus) reported 55 minutes ago

  @Saas_addy @AKirtesh @Namecheap Not when you need support it's not. Cloudflare runs a full dns scan before you set to name servers & builds the dns zone file for you. Add domain, verify in email inbox, Turn off orange clouds and change name servers at the registrar. DNS zone changes propagation auto ttl 2 min

• 
  Vlad | AI Build Lab (@VladBuildsAI) reported 57 minutes ago

  @winsontang This is a good example of why I like measuring AI workflows at the boundary, not at the model output. The source shows the visible part: "Winson Tang · 3 год Discover insights like never before with Cloudflare Radar Researcher". The useful next question is what has to be true around it for the result to survive real work. I would check 5 things: input quality, human review minutes, retry behavior, permission boundaries and recovery after the handoff from research to execution. If those are not measured, the demo can look like automation while the hidden cost sits in cleanup. The metric I would want is retry rate after the first failure. That tells you whether the workflow is reducing total work or just producing more artifacts that somebody else has to validate.

• 
  🔥🏴‍☠️CyberTechWolf🏴‍☠️🔥 (@CyberTechWolff) reported 57 minutes ago

  Cloudflare if humans are becoming a "rounding error" within 5 years why ******** do we have to verify if we're a bot on every ******* website we visit?

• 
  The ReFrame (@The_ReFrame_) reported 1 hour ago

  On the world's largest network, the traffic changed hands in May. That's the measured part. Per Cloudflare, machine traffic passed human traffic for the first time — a crossover the company had pencilled in for 2027. It came a year early. The viral number, 1,000× within five years, is something else: a forecast, conditional on trends holding. Its author attached a caveat nobody repeats — he has called it wrong at every point along the way. We've watched a network outgrow its makers before. At the end of 2009, mobile data passed voice for the first time. Voice became the rounding error — and nobody stopped calling. The network didn't abandon us; it just stopped being mainly about us. The measured story is big enough. The internet is now a place where humans are guests of honour rather than the crowd — and what we build for it is still entirely our call. #Cloudflare #AI #WebTraffic

• 
  Alexander Zuev (@zlxndr) reported 1 hour ago

  If I was looking for ways to 10x the DX of @CloudflareDev (which is already great, ofc), formulated in a withlist: 1. Make every error thrown by any binding or the runtime carry a stable, documented .code - not a code embedded in the message like R2's (10024), an actual property 2. Every such error should carry .retryable, typed, on all bindings - you already ship it on DO stubs, extend it and put it in @cloudflare/workers-types 3. Both should survive the boundaries that currently flatten errors: RPC serialization and NonRetryableError in Workflows This will enable error handling to not depend on fragile string parsing Is it too much to ask? Merci beaucoup in advance!