Claude Goes Dark: Anthropic Confirms Major Outage Across Login and Core Services
Anthropic's status page flagged authentication failures and degraded performance starting 21:58 UTC on 16 August 2026, hitting Claude.ai, Claude Code and Claude Cowork.
Key points
- Anthropic confirmed a major outage of Claude.ai, Claude Code and Claude Cowork beginning at 21:58 UTC on 16 August 2026.
- The first fault was users being unable to log in; nine minutes later Anthropic reported wider degraded performance on Claude.ai and platform.claude.com.
- The Claude API and Claude Console are still listed as operational, so apps built on the API may keep working while the consumer site does not.
- Anthropic has not said what caused the outage, and both incidents remain under investigation.
Anthropic's Claude is down for large numbers of users tonight, with the company confirming a major outage across its consumer and developer-facing products.
The first sign of trouble landed on Anthropic's public status page at 21:58 UTC on 16 August 2026. The company said it was looking into an issue that stopped some users from logging in to Claude.ai, Claude Code (its command-line coding assistant) and Claude Cowork (its team collaboration product).
Nine minutes later, at 22:07 UTC, Anthropic posted a second notice. This one was broader: degraded performance on Claude.ai and on platform.claude.com, the developer portal.
Both incidents are still open.
What are users actually seeing?
Broken logins and a chatbot that will not load. People trying to sign in are getting authentication errors, meaning the system will not accept their username and password even when the details are correct.
Others report Claude.ai failing to load in the browser, requests that never complete, and generic error screens once they are inside. First reported by BleepingComputer, the disruption is showing up across both the free and paid tiers of the consumer app.
Which services are affected?
Three products are marked as a major outage on Anthropic's status page. Two others are still listed as operational.
| Service | Status |
|---|---|
| Claude.ai | Major outage |
| Claude Code | Major outage |
| Claude Cowork | Major outage |
| platform.claude.com | Degraded performance |
| Claude API | Operational |
| Claude Console | Operational |
The split matters. If your company has built a product on top of the Claude API, that traffic should still be flowing. If your staff use Claude.ai in a browser to draft emails or summarise documents, they are likely locked out right now.
What is causing it?
Anthropic has not said. The status updates describe the symptoms (authentication problems, then degraded performance) but not the root cause. There is no mention of a cyberattack, a data exposure, or a third-party dependency failing.
Authentication outages at cloud services are usually one of a small set of causes: an expired certificate, a botched deployment, a database under load, or an identity provider (the system that checks who you are when you log in) falling over. Until Anthropic publishes a post-incident review, anything more specific is guesswork.
No regulator is involved at this stage. An outage on its own, without exposure of personal data, does not trigger notification duties under the UK GDPR, the EU GDPR or the US state breach laws. That calculus would change if Anthropic later discloses that user data was accessed or lost, but nothing in the current updates points that way.
What should users do?
Wait, mostly. There is no user-side fix for a provider-side outage. Refreshing the login page, clearing cookies or resetting your password will not help and, if the identity system is genuinely struggling, could add to the queue.
A few practical steps do make sense:
- Check status.anthropic.com directly rather than relying on social media rumours about what is broken.
- If you use Claude for work, switch to a backup workflow for anything time-sensitive tonight.
- Do not enter your Claude credentials into any "status checker" or "login fix" site that appears in search results during an outage. Fake login pages tend to surface exactly when the real one is down.
Anthropic says it is continuing to investigate. Threat Vectr will update this story when the company publishes a cause.
