Anthropic investigates authentication errors hitting Claude AI and Claude Code

Anthropic is investigating elevated authentication errors affecting Claude AI and Claude Code. Some users are unable to log in to the tools they rely on every day.

The company’s status page logged the incident as starting at approximately 09:58 UTC on November 13, 2026.

What’s broken, and where

The disruption reaches well beyond the consumer chat app. Anthropic’s status page listed claude.ai, Claude Code, and platform.claude.com among the affected services.

Anthropic described the issue as elevated errors tied to user authentication. That wording points to a problem on the company’s side, not a crowd of users who all forgot their passwords at the same moment.

Different components were classified as experiencing either a partial or a major outage. The severity varied depending on which part of the service users were trying to reach.

Users have described a range of symptoms. Reports include expired credentials, outright authentication failures, and errors tied to credential precedence and expired OAuth sessions.

OAuth is the system that lets you sign in once and stay signed in through a digital token, so you aren’t retyping a password every few minutes. Credential precedence refers to which login method a tool trusts first when it finds more than one available.

As of the latest update, Anthropic had not announced a resolution. The company has said it will keep users informed as the investigation continues.

Not the first lockout this year

This incident does not arrive in a vacuum. On September 29, 2026, Anthropic dealt with a similar episode involving elevated authentication errors and sign-in issues.

The recurring nature of these problems has historically produced widespread confusion and frustration among users, who often can’t tell whether the fault lies with their own setup or with Anthropic.

Earlier in 2026, session-token theft incidents were reported, tied to infostealer malware. Those incidents led to unauthorized usage of accounts and to account suspensions.

Infostealers are malicious programs that sit quietly on a victim’s computer and copy saved logins and session data. A stolen session token works a lot like a valet ticket: whoever holds it can drive off with the car, and nobody asks for ID.

The available reports do not tie the current errors to that theft. One involves outsiders getting in where they shouldn’t. The other involves legitimate users being kept out of accounts they own.

What this means for Claude users and Anthropic

For individual users, an authentication outage is a particularly maddening kind of failure. A capable AI model does you no good if you can’t get past the login screen to use it.

With Claude Code and platform.claude.com both in the affected list, people who have built Anthropic’s tools into their daily workflows could see that work grind to a halt until access is restored.

Anthropic’s AI tools are seeing greater adoption, which means more people are depending on them. Continued service disruptions could deter new users and affect subscription rates, an important metric for measuring the success of their growing platform.