Anthropic’s Claude AI platform ran into significant trouble Monday morning, with thousands of users across the globe reporting access problems that grew steadily throughout the early hours of April 6. What began as a scattered wave of complaints escalated quickly, ultimately drawing more than 8,000 user reports on Downdetector, the crowdsourced outage tracking platform.
How the outage unfolded
The disruption built in stages throughout the morning. By 8:26 a.m. PDT, more than 2,500 users had already flagged problems on Downdetector, with the majority citing issues specifically with Claude Chat. Within roughly 15 minutes, that number had climbed past 3,300. By 8:40 a.m., nearly 5,000 reports had come in, and by the top of the hour the count had surpassed 8,000 — a sharp escalation that pointed to a problem affecting a broad cross-section of the platform’s user base rather than an isolated technical hiccup.
For much of the early morning, Anthropic‘s own status checker continued to display an “All Systems Operational” message, creating a disconnect between what users were experiencing and what official monitoring was indicating. That changed at 8:48 a.m. PDT, when the status page was updated to acknowledge an active incident. The company confirmed elevated errors across Claude.ai on both desktop and mobile surfaces, with users encountering problems when trying to log in, use voice mode or complete chat sessions. A subsequent update extended the scope of the incident to include login issues on Claude Code as well.
What users were experiencing
The complaints that surfaced on Downdetector and across social media platforms painted a fairly consistent picture. Login failures were among the most widely reported issues, particularly affecting Claude Code and the desktop application. Users attempting basic web chat through claude.ai reported mixed results — some were able to access the interface intermittently while others found it unresponsive. Voice mode and multi-turn conversation completion were also disrupted for a meaningful portion of those affected.
Developers who rely on Claude Code for daily workflows were among the most vocal about the disruption, with some reporting that repeated authentication errors forced them to switch to alternative tools mid-session. The timing — overlapping with typical workday peak hours across multiple time zones — amplified the frustration for professionals who depend on the platform for coding, writing and research tasks.
Context: a pattern of growing pains
Monday’s incident did not emerge in isolation. March 2026 had already been a challenging month for Anthropic’s infrastructure, including a major outage on March 2 that left claude.ai and its associated applications inaccessible for several hours, and a separate five-hour disruption later in the month that affected chat and app access for users worldwide. Elevated error rates tied to specific models, login path failures and capacity strain had surfaced on several other dates throughout the month as well.
Uptime figures for early April had shown some improvement over March’s 98.21% availability rate for claude.ai, with the month tracking closer to 99.43% before Monday’s incident. The platform’s API has generally maintained stronger reliability than its consumer-facing surfaces, making it a practical fallback for developers and enterprise users during periods of instability.
What to do if you are still affected
For users continuing to experience problems, Anthropic recommends checking the status page at status.claude.com for real-time updates. Standard troubleshooting steps include refreshing the browser or app, clearing cache and cookies, trying an alternative login method such as Google sign-in if email-based authentication is failing, or checking for available desktop app updates. Developers with API access may find that surface more stable during consumer-side disruptions. Paid subscribers who experience unusual service limitations during a confirmed incident can contact Anthropic support to discuss potential account credits.
Anthropic has historically moved quickly to resolve incidents once root causes are identified, typically following up with infrastructure adjustments and, in significant cases, post-incident documentation for affected API users.
Sources: IBTimes AU, Heavy.com