Around 22:00 UTC on August 16, things just stopped for a lot of people. Anyone trying to sign in to Claude.ai, Claude Code, or Cowork couldn’t get through. Shortly after, it widened into a general disruption.
What happened
Anthropic’s status page first reported an authentication problem: some users couldn’t log in to Claude.ai, Claude Code, and Cowork. A broader notice followed – degraded performance on Claude.ai and platform.claude.com. In the end five services were affected, including the API that carries enterprise and third-party workloads.
The fix came fast. Around 22:22 UTC Anthropic said it had deployed a correction. By 22:40 UTC all services were back. That’s roughly 36 minutes – annoying, but short.
No cause given
What’s missing is an explanation. So far Anthropic has published no root cause and no postmortem. For an outage that starts at sign-in and then spreads to several services, it sounds like a shared building block – auth is the kind of spot where a single fault hits everything at once. But that’s speculation until something official lands.
My take
A half-hour outage on its own isn’t a big deal. Servers fail, that’s part of it. What’s more interesting is what it shows: how many paths now run through the same door. Claude.ai, Code, Cowork, the API – if sign-in hangs, all of it hangs.
If you run Claude in production, keep that in mind. A fallback for when the login jams isn’t distrust of Anthropic, it’s just basic operational hygiene. And I’d still like a short, open postmortem – not because of these 36 minutes, but because that’s exactly where trust comes from.
Sources: