Anthropic confirmed elevated error rates across multiple Claude models on June 23, 2026, hitting Claude.ai, the Claude API, Claude Code, and Co work.
Down detector reports peaked near 7,119 in the US before Anthropic pushed a fix and marked the incident resolved.
The outage is the latest in a run of June disruptions, landing as Anthropic moves toward a closely watched IPO.
Claude went down for thousands of users on June 23, 2026, and Anthropic confirmed an elevated error rate while it scrambled to restore service. The company's status page logged the issue as investigating at 14:19 UTC, marked it identified by 14:25 UTC, and began rolling out a fix soon after. The disruption touched the full Claude lineup, including Claude.ai, Claude Console, the API, Claude Code, and Co work.
What Happened
User reports spiked on Down detector starting around 10:02am ET, peaking at 7,119 in the US before dropping back to normal. Anthropic confirmed the elevated error rate across multiple models and said it had resolved the incident by 12:44pm ET.
Claude for Government stayed online throughout. Outage tracker Status Gator logged nearly 1,300 user reports over 24 hours before the service recovered.
This was not a one-off. June has brought a steady run of Claude incidents, including a major multi-hour outage on June 2 and a roughly 90-minute multi-model disruption on June 22.
Anthropic told Fortune earlier this year that demand has grown faster than its infrastructure can handle at peak hours, and it is adding capacity through partnerships with Amazon and Google.
The official Anthropic account and live user reports tracked the issue here:
https://x.com/search?q=%23claudedown
What To Do If Claude Is Not Working
Check status.claude.com for the live incident status first. If the page shows an active incident, the problem is on Anthropic's side and waiting is usually the fix. You can also retry your request, clear your browser cache, switch between desktop and mobile, or try a VPN.
Developers seeing HTTP 500 errors are hitting a server fault, while HTTP 529 means the system is overloaded and rejecting requests until capacity frees up.
This tech news recap covered the June 23 outage:
https://www.youtube.com/watch?v=nnxR8aed4RY
Why It Matters
The timing is awkward. Anthropic filed a confidential draft IPO prospectus with the SEC on June 1, 2026, after a $65 billion funding round at a $965 billion valuation, with a public listing widely expected later this year. A run of visible outages tests user trust at the exact moment the company is courting public market investors.
https://www.youtube.com/watch?v=E-vnEbJnYz8
Are AI services reliable enough for critical business use? For now, the smart move is a backup plan that does not depend on a single provider.
Quick Links:
