On Monday, June 23, Claude had a major outage. Problems started around 10 AM Eastern Time, and for roughly 85 minutes the service was disrupted across the board. Affected: claude.ai, the API, Claude Code, Console, and Cowork — essentially everything Anthropic offers.
What happened
Anthropic’s status page confirmed an ‘elevated error rate’ across all models. At the peak, over 7,000 users in the US reported issues simultaneously. The incident was marked as resolved at 12:44 PM ET — about two hours after the first alert.
There’s been no official explanation of what actually went wrong.
One notable detail
If you’re on Claude for Government, you didn’t notice a thing. The Government instance ran without interruption throughout. That makes sense — it’s separate infrastructure with higher resilience guarantees — but it still raises questions. If the same technology runs fine in one environment but fails in another, the problem clearly isn’t the model itself.
Context
This wasn’t the first recent outage. Just three weeks earlier, on June 2, there were also problems. This one was significantly larger and more widespread, though.
For developers building on Claude’s API, these outages are a real concern. 85 minutes doesn’t sound like much, but when your product depends on Claude and error rates spike, you feel it immediately.
My take
Outages happen — to every cloud service. What bothers me is the lack of transparency. No explanation, no post-mortem, just a ‘resolved’ on the status page. Anthropic should communicate more openly here, especially if they want to win enterprise customers who depend on reliability. A brief ‘here’s what happened, here’s what we did’ would be a good start.
Sources: