⚠ IS CLAUDE AI DOWN? — CLAUDE GOES DOWN MULTIPLE TIMES PER WEEK     137 INCIDENTS IN 90 DAYS     ANTHROPIC WILL NOT NOTIFY YOU     CHECK STATUS BELOW     ⚠ IS CLAUDE AI DOWN? — CLAUDE GOES DOWN MULTIPLE TIMES PER WEEK    
⚠ Claude Status — Updated April 28, 2026

Is ClaudeAi Down?

If you're getting errors, timeouts, or login failures on Claude.ai — it's probably not your connection. Claude goes down multiple times per week. Here's how to verify it, what the error codes mean, and what you can do about it.

// where to check right now
Official Source
STATUS.CLAUDE.COM
Anthropic's official status page. Updated during incidents but typically 15–30 minutes behind actual outage start. Only covers what Anthropic chooses to acknowledge. 87 incidents in March 2026 alone were never reported here.
Check status.claude.com →
Third-Party Monitor
STATUSGATOR
Detects outages up to 2.3 hours before Anthropic acknowledges them. Tracks all 6 Claude components independently. More reliable early warning than Anthropic's own page. Free tier available.
Check StatusGator →
User Reports
DOWNDETECTOR
Crowdsourced user reports. Good for confirming you're not alone. During the April 28, 2026 outage, over 12,000 users reported issues here before Anthropic issued a resolution notice.
Check Downdetector →
// what the errors actually mean

Error Codes
Decoded.

Claude produces specific error codes during outages. Here is what each one means and whether it is Anthropic's problem or yours.

Http 500
Internal Server Error
Anthropic's problem. Their servers are returning an unhandled error. Nothing you can do on your end will fix this. Check status.claude.com. Wait. If you are a Pro subscriber, document the time and date for a refund request.
Http 529
Overloaded
Anthropic's problem. Their servers are receiving more requests than they can handle. This is a capacity failure, not a user error. The 529 code appeared extensively during both the April 28 and April 15, 2026 outages. Document and request a refund.
"Service is
temporarily busy"
Anthropic's problem. Generic message displayed during outages when Claude's frontend is loading but the backend is unavailable. The page appears functional but no prompts will process. Confirmed during the April 15 outage by TechRadar's live coverage.
"Authentication
service unavailable"
Anthropic's problem. Login paths are broken. You cannot authenticate even if the underlying model is operational. Pro users are disproportionately affected because their accounts require verification codes that the broken authentication service cannot process.
"You've used your
free messages"
Probably Anthropic's problem. During the April 15 outage, free users were falsely told they had exhausted their daily message limit despite having sent zero messages. If you see this unexpectedly, check Downdetector before assuming it is accurate.
"Can't reach Claude
— check your connection"
Could be either. Check your internet connection first — a 10-second test. If your connection is fine and you can reach other sites, go directly to status.claude.com. This message appears during full outages and is designed to suggest the problem is on your end when it frequently is not.
// what to do when claude is down

Step By Step:
When It Fails.

01
Verify It's Not Your Connection
Open any other website. If they load normally, your connection is fine and the problem is Anthropic's. This takes 10 seconds and eliminates the most common false diagnosis.
02
Check The Official Status Page
Go to status.claude.com. If it shows an active incident, Anthropic has confirmed the problem. If it shows all systems operational but you still cannot connect, the outage may not yet be acknowledged — this happens regularly. StatusGator detects outages up to 2.3 hours before Anthropic admits them.
03
Check Downdetector For User Reports
Go to downdetector.com/status/claude. A spike in reports confirms you are not alone even if the official status page has not caught up. During major outages, Downdetector reaches thousands of reports before Anthropic acknowledges the incident.
04
Document The Outage Immediately
Note the exact time you first experienced the failure, the error message displayed, your subscription type, and any work that was interrupted. This documentation is required if you want to request a refund. Anthropic does not track your outage experience for you and will not proactively offer compensation.
05
Wait — Median Resolution Is 68 Minutes
Based on third-party monitoring data covering 137 incidents in 90 days, the median outage duration is 68 minutes. Major outages can run significantly longer — the March 2–3, 2026 incident spanned nearly 20 hours across three waves. There is no mechanism to expedite resolution. You wait or you leave.
06
File For A Refund After
Anthropic will not contact you. You must submit to support@anthropic.com with your outage documentation and a specific refund request. See our refund guide for the exact language that gets results.
// how often does this actually happen

How Often
Claude Goes Down.

This is not a product that occasionally has issues. This is a product that goes down multiple times per week as a matter of documented record.

965+
Total incidents since
June 2024
2+
Average incidents
per day
43
Major outages in
last 90 days
68min
Median time to
resolve an outage
// most recent confirmed outages
Apr 28, 2026 MAJOR Claude.ai unavailable and elevated API errors — 1hr 24min — 12,000+ user reports
Apr 25, 2026 WARNING Degraded service globally — 18hr 43min — North America, Europe, Asia-Pacific affected
Apr 24, 2026 MAJOR Claude.ai down — 39min — preceded by 13hr 50min warning period same day
Apr 20, 2026 MAJOR Claude.ai down — 6hr 50min — no proactive subscriber notification
Apr 15–16, 2026 MAJOR Claude.ai and Platform down — ~3hr — 30,000+ user reports — login completely broken
Mar 2–3, 2026 MAJOR Worst outage on record — ~20hr across 3 waves — 4,700 Downdetector reports at peak
// the honest context

What This
Actually Means.

The thing Anthropic won't say
If Claude is down right now and you searched for this page, you are experiencing one of the most predictable events in AI — Claude failing. It happens multiple times per week. The median resolution time is over an hour. The worst outages run for most of a day.

Anthropic will not email you when it happens. They will not text you. They will not proactively offer you a refund. Their status page typically lags 15 to 30 minutes behind the actual start of an incident. Third-party monitors detect outages up to 2.3 hours before Anthropic acknowledges them.

You are paying for a service with no SLA, no uptime guarantee, no proactive communications, and no automatic compensation. Every time you Google "is Claude AI down" and find this page, that is not a random bad day. That is the product operating as it normally operates.

You have the right to demand a refund for every billing cycle where this has materially impacted your use of the service. Most subscribers don't. That is exactly how Anthropic has structured this.

It's Down.
Now What.

While you wait for Claude to come back up — or decide it's not worth waiting for — here's what you can actually do.

// demand a refund

Document this outage. Note the time, date, error message, and subscription type. Submit to Anthropic's support with a specific refund request. They will not offer it unprompted.

support@anthropic.com usersafety@anthropic.com

// submit your experience

Add this outage to the public record at ClaudeSucks. Your documentation helps other subscribers understand they are not alone and builds the evidentiary record.

claudesucks.workwithvsg.com

// escalate to legal

Repeated outages with no SLA and no automatic compensation is a consumer protection issue. Jeffrey Bleich is General Counsel. Certified mail creates a record.

Anthropic PBC, 548 Market St
PMB 90375, SF CA 94104
Attn: Jeffrey Bleich, GC