Claude sign-ins fail on claude.ai and Claude Code, and a fix is still being watched

Anthropic’s status page reports elevated errors on claude.ai, Claude Code and platform.claude.com on November 13, 2026. The company describes the cause as users failing to authenticate. A fix was deployed at 11:05 UTC and the incident remains open.

A sign-in problem that spread across four products

According to the incident page, Anthropic first posted an “Investigating” notice at 09:58 UTC. It said users were having trouble authenticating to Claude.ai and Claude Code, and that more updates would follow. At 11:05 UTC the status moved to “Monitoring”, with a note that a fix had been deployed and the team was watching for recovery.

The page lists four affected components: claude.ai, the Claude Console at platform.claude.com, Claude Code and Claude Cowork. The incident title also mentions “elevated errors”, but the text describes only authentication failures. It gives no error messages, error rates or codes.

For a reader, the practical effect is that a login problem can block work even when the models behind the products are healthy. Anyone who relies on Claude Code in a terminal, or on the web app, could be unable to start or resume a session until authentication works again. The page does not say whether users who were already signed in were affected, or whether API traffic was touched.

What the status page does not tell us

The incident page leaves most of the important questions open. It does not name a cause. It does not say how many users were affected, how long individual users were locked out, or which regions saw the problem. It also does not say whether the fix was a rollback, a configuration change or a repair of an upstream dependency. The wording “a fix has been deployed” is Anthropic’s own claim, and recovery had not been confirmed when we checked.

The overview page was also not fully consistent with the incident text when we read it. It listed claude.ai as having a major outage, while the Claude Console, the API at api.anthropic.com and Claude Code were shown as operational. That is a snapshot and may lag behind the incident page, which names Claude Code and the Console as affected. It does suggest the impact was uneven across products, and that the API, which developers use directly with keys, may have been spared. Anthropic has not said so explicitly.

A post-incident write-up, if one comes, would be the place to learn the root cause. None had been published at the time of writing.

One of several recent incidents on the same page

The same status page lists other recent entries. On November 12 it flagged elevated errors affecting the Microsoft 365 connector, and its latest update there (22:23 UTC) attributes the problem to a Microsoft 365 service incident. On November 10 it recorded elevated errors on Fable 52 and Mythos 52, with overloaded errors lasting four minutes. On November 8 it logged degraded performance for Claude Fable 5.1 from 08:18 to 08:32 UTC, about 14 minutes.

Against those, the November 13 incident is the broadest in scope: it touches sign-in rather than a single model or connector, and it had already run for more than an hour when the fix was announced. The earlier model incidents were measured in minutes.

Rivals publish similar notices. Our earlier report on OpenAI’s status page flagging a wide outage on October 21 described 33 affected components across ChatGPT, Codex and the API, and noted that OpenAI gave no cause. The same page listed an October 13 incident that stayed open for roughly 8 hours 18 minutes after a mitigation was applied. Both companies tend to give short, cause-free updates while an incident is live, so users usually have to wait for any later explanation.

The pattern matters for teams that have built daily workflows on a single assistant. A login failure is hard to work around, because there is no fallback inside the product. The status page remains the only official source for now, and its latest word is that recovery is being monitored, not that the incident is over.