Anthropic’s Claude AI suffered a confirmed global outage early Tuesday morning, taking down the chat interface, API, Console, and Claude Code simultaneously for a stretch of roughly 35 minutes. The root cause was identified and a fix rolled out before most of the US woke up — but not before “is Claude down” became one of the top trending searches in both the US and UK.
Claude AI went down on June 2, 2026 due to a confirmed Anthropic infrastructure incident. The outage began around 2:19 AM ET (06:04 UTC), affected Claude.ai, the Anthropic API, Console, and Claude Code, and was fully resolved by approximately 06:39 UTC. Anthropic first confirmed it was investigating elevated error rates, then identified the root cause and rolled out a fix within 35 minutes.
What Was Affected
The incident wasn’t limited to casual users. Anthropic’s status page confirmed elevated error rates across four services: Claude.ai (web and mobile), the Anthropic API, the developer Console, and Claude Code — the agentic coding CLI that has quietly become Anthropic’s fastest-growing product. Developers building on Claude hit failed requests in the early hours; users on the free and paid tiers saw error messages or unresponsive sessions. Billing and subscription management on claude.ai and the Console also showed errors for some users, though Anthropic clarified that chat and API access were the primary concern.
The outage drew particular attention because of the timing: Anthropic is in the middle of a high-profile stretch, with its confidential IPO S-1 filed just days ago and its Mythos-class model — capable of expert-level autonomous hacking — on the verge of a public release. Outages at this juncture tend to amplify scrutiny.
How Quickly Was It Fixed?
Anthropic’s incident response was fast by industry standards. The company moved from “investigating” to “root cause identified” to “fix being implemented” in under 40 minutes. By 3:48 AM ET, StatusGator was showing all Anthropic services as operational. For context, the 2024 OpenAI ChatGPT outage that hit the US East Coast during peak hours lasted nearly three hours — Anthropic’s window here was substantially shorter, and the impact landed in off-peak hours for its largest user base.
Our Take
A 35-minute outage resolved before the US workday starts is not a crisis — it’s infrastructure. What matters is the pattern. Claude has grown fast enough that any blip now moves search trends in two countries simultaneously. That’s a sign of genuine adoption, and it’s why reliability engineering matters more to Anthropic today than it did 18 months ago. If you’re a developer building on the API, this is a good reminder to wire up status.claude.com alerts and build retry logic. If you’re a Claude Max subscriber who was mid-session, the fix is live — you’re good. Anthropic has not published a full post-mortem yet; we’ll update if one appears.
