The outage hit Claude.ai, Claude Code and the API simultaneously before services were restored
For a stretch of nearly 80 minutes this afternoon, one of the most widely used artificial intelligence platforms in the world went dark for a significant portion of its user base, and the numbers that piled up on outage tracking sites told the story clearly.
Claude AI, Anthropic’s flagship AI platform, experienced a major service disruption on April 28, 2026, knocking out access to Claude.ai, Claude Code, Claude Chat and the Anthropic API simultaneously. By the time the incident reached its peak, more than 12,000 users had filed reports on Downdetector, and Anthropic’s own status page had escalated the event to a full Major Outage classification.
How the outage unfolded
The first wave of trouble surfaced around 10:59 a.m. PT, when reports of access failures began appearing on Downdetector. Users attempting to log into Claude.ai were met with authentication errors, while those working through Claude Code and the API encountered elevated failure rates that disrupted workflows mid-session. The most commonly flagged issues centered on Claude Chat and Claude Code, two of the platform’s most heavily used interfaces.
Within roughly 30 minutes, the report count had climbed past 5,000. By 11:31 a.m. PT, it had nearly doubled again to close to 8,000. At 11:53 a.m. PT, with more than 12,000 reports logged, Anthropic confirmed on its status page that the company was continuing to work to resolve issues preventing users from accessing Claude.ai and causing elevated authentication errors across the API and Claude Code.
The outage window, according to Anthropic’s own incident data, ran from 17:34 UTC to 18:52 UTC, a span of approximately 78 minutes. By 12:02 p.m. PT, the company reported that success rates across all services had returned to normal and that engineers were monitoring closely to prevent any further disruption.
A second major incident in just 8 days
What made the April 28 outage particularly notable was that it was not an isolated event. Just 8 days earlier, on April 20, 2026, Claude AI had suffered a separate major disruption that followed a similar pattern. During that incident, Downdetector recorded 6,500 users reporting problems across Claude Chat, Claude Code and the Claude mobile app. Anthropic classified that event as a Partial Outage tied to authentication failures, with the API recovering first before the web interface and mobile access stabilized.
The April 20 incident itself had come on the heels of another disruption earlier in the month, making the back-to-back failures a growing point of concern among enterprise users and developers who route critical workflows through Claude’s infrastructure.
Why reliability concerns are growing
For individual users, a service interruption lasting a little over an hour is an inconvenience. For the growing number of businesses and development teams that have embedded Claude into their daily operations, even brief outages carry real operational costs. As enterprise adoption of AI tools accelerates across industries, the reliability of the underlying infrastructure has become as important as the quality of the model itself.
Anthropic has not yet published a detailed postmortem for either the April 20 or April 28 incidents, and the company has not offered a public timeline for infrastructure improvements aimed at preventing further recurrences. Whether the repeated failures prompt a more formal transparency effort from the company remains to be seen, but the volume of user reports and the back-to-back nature of the events suggest the pressure to respond is only growing.