Claude outages on November 13 hit logins, Claude Code and connectors for 4.5 hours
Updated: 15 minutes ago
Claude had two separate service incidents on Friday, November 13, 2026, and together they kept parts of claude.ai, Claude Code and Anthropic's connector ecosystem impaired for 4 hours and 29 minutes of a single working day. The first incident, posted on Anthropic's official status page at 09:58 UTC, was an authentication failure that stopped many users from signing in to claude.ai and Claude Code; it was marked resolved at 11:58 UTC. The second began at 13:09 UTC as degraded performance on claude.ai, the Claude API and Claude Code, widened over the following hour to connectors, Claude Managed Agents, Claude Security and Claude in Meetings, and was marked resolved at 15:38 UTC.
The two incidents overlapped with a third, longer problem that had started the previous evening. From 17:52 UTC on November 12, Anthropic's status page tracked elevated errors in the Microsoft 365 connector, which left mail, calendar and search tools failing for most requests and also affected Claude for Outlook. Anthropic attributed that incident to a Microsoft 365 service incident, and by 15:40 UTC on November 13 it had moved to monitoring with signs of recovery, while some SharePoint search requests were still returning errors. For users in the United States, the morning incident began at 4:58 a.m. Eastern and the afternoon incident ran from 8:09 a.m. to 10:38 a.m. Eastern, squarely inside the start of the US workday.
··········
CLAUDE STATUS PAGE TIMELINE FOR NOVEMBER 12 AND 13.
The table below reconstructs the three incidents from the update timestamps on status.claude.com. All times are UTC; the durations are calculated from the first "Investigating" post to the "Resolved" post, or to the latest update where no resolution has been posted.
........
Incident | First post (UTC) | Fix or recovery | Resolved (UTC) | Duration |
Authentication errors on claude.ai and Claude Code | Nov 13, 09:58 | Fix deployed 11:05 | Nov 13, 11:58 | 2 h 00 min |
Elevated errors on claude.ai, the API and Claude Code | Nov 13, 13:09 | Recovery seen 14:36 | Nov 13, 15:38 | 2 h 29 min |
Microsoft 365 connector errors | Nov 12, 17:52 | Recovery seen Nov 13, 15:40 | Monitoring, not yet resolved | 21 h 48 min to monitoring |
........
The morning authentication incident moved quickly from investigation to a fix. Anthropic's first post said it was investigating "an issue with users authenticating to Claude.ai and Claude Code", a deployed fix was announced 67 minutes later, and the company then monitored for close to an hour before marking the issue affecting claude.ai and Claude Cowork as resolved. The afternoon incident followed a different pattern: the scope grew with each update, from the three core surfaces at 13:09 UTC to connectors such as Docs and Sheets at 13:52 UTC and then to Managed Agents and Claude Security at 14:03 UTC.
··········
WHICH CLAUDE PRODUCTS WERE AFFECTED BY THE NOVEMBER 13 INCIDENTS.
During the morning incident the component board on the status page did not show the same impact for every product. The levels below are those displayed for each component while the authentication problem was active.
........
Claude component | Status during the morning incident |
claude.ai | Major outage |
Claude Code | Partial outage |
Claude Console (platform.claude.com) | Partial outage |
Claude Cowork | Degraded performance |
Claude API (api.anthropic.com) | Operational |
........
The split between a major outage on claude.ai and an operational API is consistent with the nature of the problem, which Anthropic described as a sign-in failure. Applications calling the API with keys did not depend on the interactive login flow, so production integrations built on the API were the least exposed part of the platform that morning. Developers who use Claude Code through a claude.ai subscription login, and teams administering accounts in the Console, were affected.
The afternoon incident was broader and did reach the API. Anthropic's update at 14:36 UTC reported recovery on claude.ai and the API while warning that errors could recur, and said that connectors, Claude in Meetings, Managed Agents, Claude Security and the use of connectors inside Claude Code remained degraded. The 15:33 UTC update added the mobile apps to the list of surfaces whose error rates had returned to normal. For enterprise teams that have adopted the newer agent products, including the self-hosted option covered in our report on self-hosted Claude Managed Agents, and the Google Meet integration described in our article on Claude Meetings, the second incident was the more disruptive one.
··········
[[ADPLUS_300x250_1]]
WHAT CHANGES FOR CLAUDE USERS AFTER THE OUTAGE.
The resolution posts left several after-effects that users need to check for themselves. At 15:33 UTC Anthropic said that connectors were working again but that adding a new connector on claude.ai was temporarily unavailable. The same update warned that scheduled tasks and routines might run more than two hours late and that some runs had been skipped entirely. A skipped run does not repeat on its own schedule, so anyone relying on Claude for recurring reports, monitoring jobs or automated summaries on November 13 should open the task history, confirm which runs completed, and trigger the missing ones manually.
Users of the Microsoft 365 connector and Claude for Outlook were dealing with a separate dependency. Anthropic's updates explained that mail, calendar and search tools were failing for most requests because of a Microsoft 365 service incident, that some users could not open the Outlook add-in, and that progress could be followed on Microsoft's own status page. At the latest update, recovery was under way and Microsoft's incident remained open, so organizations that route Outlook mail, Teams chat or SharePoint content through Claude should verify the connector before relying on it for time-sensitive work.
For readers comparing plans, an outage of this kind does not change the usage limits, prices or plan features described in our guide to Claude pricing and plan limits. Anthropic had not announced any credit or compensation for the November 13 incidents at the time of writing. The practical steps for individual users and administrators are straightforward:
........
User group | What to check after November 13 |
claude.ai and mobile users | Sign in again if a session was dropped; retry conversations that failed mid-response |
Claude Code users | Re-authenticate if prompted; re-run jobs that used connectors between 13:09 and 15:38 UTC |
Scheduled tasks and routines | Review run history for late or skipped runs and trigger missed ones manually |
Connector administrators | Retry adding new connectors later; confirm Docs and Sheets connectors respond |
Microsoft 365 and Outlook users | Check Anthropic's and Microsoft's status pages before relying on mail, calendar or SharePoint search |
API developers | Review error logs for the 13:09 to 15:38 UTC window; the API was listed as operational during the morning incident |
........
Subscribing to updates on status.claude.com is the quickest way to learn about a new incident, and teams that run production workloads on the API can add retry logic with backoff for the class of elevated-error events seen on the afternoon of November 13.
··········
CLAUDE RELIABILITY ON NOVEMBER 13: CONFIRMED FACTS AND OPEN QUESTIONS.
The official record establishes the times, the affected surfaces and the resolution of both Friday incidents, along with the cause that Anthropic assigned to the Microsoft 365 connector problem. Anthropic has not published a root cause for the authentication failure or for the afternoon errors, has not said what share of users or requests were affected, and has not said whether the two incidents were related. The status page also offers no impact figures beyond the per-component labels.
One further detail remained unexplained when this report was verified late on November 13 UTC. The component board still showed claude.ai and Claude Cowork with degraded performance, while every other component, including the Claude API, Claude Code and the Console, was listed as operational, and no new incident post accompanied those labels. Until Anthropic publishes a post-incident review, the established picture is two resolved incidents totaling 4 hours and 29 minutes on November 13, an API that stayed up during the login failure but not during the afternoon errors, and a Microsoft-caused connector problem that was still in monitoring.
··········
FOLLOW US FOR MORE.
·····
DATA STUDIOS
·····
datastudios.org
