Claude AI experienced another significant service disruption on April 28, 2026, marking the third major outage in less than a month. The Claude AI outage began around 8:30 AM Central Daylight Time, with users reporting slowdowns, login failures, and billing-related errors across web, mobile, and API services.
Key Takeaways
- Claude AI outage on April 28 started around 8:30 AM CDT with billing errors and service slowdowns
- Anthropic’s status page showed elevated billing errors resolved between 12:24–12:45 UTC, but users reported ongoing issues
- This is the third major disruption in three weeks; April 15 outage affected 15,000–20,000 users globally
- Users experienced API Error 500, login failures, and false “service busy” messages despite official resolution claims
- Downdetector reports lag behind actual recovery, with users continuing to experience slowdowns after official “fixed” announcements
Claude AI outage timeline: What happened on April 28
The Claude AI outage began with reports of severe slowdowns around 8:30 AM Central time on April 28. Users on Downdetector described the service as “really really really slowing down” before complete disruption set in. The outage affected multiple layers: web access, mobile apps, API endpoints returning Error 500, and billing systems that generated false error messages. Anthropic’s status page logged elevated billing-related errors and applied a fix between 12:24 and 12:45 UTC. However, user reports suggest the official resolution timeline does not match real-world experience — some users received email notifications claiming the issue was fixed but continued experiencing slowdowns.
The pattern mirrors earlier disruptions. On April 15, a more severe outage began around 2:53 PM UTC with errors on Claude.ai and the Claude Code platform, escalating to full downtime by 3:40 PM UTC. That outage spiked Downdetector reports to between 15,000 and 20,000 affected users before resolution at 5:42 PM UTC. A week earlier, on April 6–7, login, voice mode, and chat functions all failed, with Downdetector peaking at 2,900 reports.
Who was affected by the Claude AI outage
The Claude AI outage disrupted service across all user tiers and access methods. Free-tier users faced false “prompts used up” errors during earlier outages, unable to distinguish between genuine rate limits and service failures. Pro subscribers experienced login verification failures that locked them out entirely. API users received Error 500 responses, breaking integrations and automated workflows. Claude Code users, who access the feature through Claude.ai, could not log in despite existing sessions remaining functional during the April 15 event. The global nature of the outages meant disruptions spanned regions simultaneously, with no geographic carve-outs reported.
Why the status page and real experience diverge
A persistent tension emerged during the April 28 Claude AI outage: Anthropic’s official status page claimed resolution while Downdetector and user reports showed ongoing problems. One user stated bluntly: “Claude is down regardless of their status page, it lies”. Another reported receiving an official email confirming the fix but continuing to experience slowness. This gap reflects a common infrastructure reality — status pages track when engineers apply fixes and success rates begin recovering, but full propagation across global CDNs and cached states takes longer. Users hitting regional servers still experiencing congestion see the outage continuing even after the status page declares monitoring complete. The April 15 outage showed similar delays, with Downdetector reports remaining elevated after the status page shifted to “Monitoring”.
Recurring outages raise reliability questions
Three major disruptions in 21 days suggests Claude’s infrastructure is under strain. The April 15 outage was the most severe, with full platform downtime affecting login verification, prompt counting, and global access. The April 6–7 incident disrupted login, voice, and chat across desktop and mobile. The April 28 event focused on billing systems but still triggered slowdowns and login issues. Each outage used Anthropic’s standard recovery phases — Identified (issue pinpointed), Monitoring (fix applied, success rates recovering), and Resolved — but the frequency indicates either rapid feature deployment without sufficient capacity planning or infrastructure scaling that has not kept pace with user growth. Anthropic has not disclosed the root causes of these outages publicly, leaving users to speculate whether the issues stem from database bottlenecks, load balancer failures, or third-party dependency problems like Cloudflare (which one user mentioned as a contributing factor before it was resolved).
What happens to your data during a Claude AI outage
During the Claude AI outage, existing chat sessions remained accessible for users already logged in, but new login attempts failed. This suggests Anthropic’s authentication layer separated from the primary service layer — a common architecture that protects data but creates a partial-outage experience. Billing errors during the April 28 event did not result in accidental charges or data loss according to available reports, though the false error messages created confusion about account status. The outages did not appear to corrupt conversation history or user settings, as users could resume work once services recovered. However, integrations relying on API availability experienced interruptions, potentially causing downstream failures in applications that depend on Claude.
Frequently asked questions
Is Claude AI still down on April 28?
Anthropic’s status page showed the April 28 Claude AI outage resolved at 12:45 UTC. However, user reports on Downdetector indicated slowdowns and access issues persisted beyond the official resolution window. Check the live status page at status.claude.com for current status, but expect real-world recovery to lag behind official announcements.
How often does Claude experience outages?
Claude has experienced at least three major service disruptions in the past three weeks: April 6–7, April 15, and April 28, 2026. This frequency is unusually high for an AI service at Claude’s scale and suggests infrastructure challenges that Anthropic is actively working to resolve.
Should I switch to another AI service because of Claude outages?
Outage frequency is one factor in choosing an AI service, but Claude’s underlying capabilities remain competitive. If your workflow cannot tolerate service disruptions, consider using Claude as a secondary tool alongside another provider. If you rely on Claude API integrations, implement fallback logic or retry mechanisms to handle temporary failures gracefully.
The Claude AI outage on April 28 is the latest signal that rapid growth is straining Anthropic’s infrastructure. Three major disruptions in three weeks is not normal for a mature service, and the gap between official status updates and user experience suggests communication and monitoring need improvement. For now, users should assume Claude may be unavailable without warning and plan accordingly.
This article was written with AI assistance and editorially reviewed.
Source: Tom's Guide
