If Claude asked you to log in again on Friday afternoon and then refused the login, it was not your Wi-Fi and it was not a bad password. On 13 November 2026 Anthropic posted on its status page at 9:58 UTC, which is 3:28 pm IST, that it was investigating elevated authentication errors for people trying to sign in to Claude.ai and Claude Code. Moneycontrol published that confirmation at 17:54 IST the same day. Business Today’s desk in Noida updated its version at 6:05 pm IST. Neither story carries a restore clock. Anthropic said it would share more when it had an update, and it did not give an estimated time.
That is the message Indian developers were staring at while a Microsoft 365 outage was already eating Outlook and Teams. Two different failures, same Friday. If your chatbot tab and your office mail both died, do not mash them into one incident. Anthropic has not linked them.
What was actually down
The status dashboard did not paint every Claude product the same colour. Both desks report the same split.
- Claude.ai was marked as a major outage. That is the chat most people open in a browser.
- Claude Code was under a partial outage. That is the coding tool. A partial tag means some sessions worked and some did not. It is not an all-clear.
- Claude Console, at platform.claude.com, was also a partial outage. Console is the developer platform, so billing, keys and project settings sat in the same broken login path.
- Claude Cowork was reporting degraded performance. Degraded is a third bucket, below a partial outage, and the articles do not spell out which Cowork actions failed.
- Claude API stayed listed as operational. Scripts and products that call the API directly were not in this login failure.
The practical split is the one to remember. A website login and a developer-console login broke. An API key that was already issued kept working, as far as the status page told either newsroom. If your app answered and claude.ai would not let you in, that matches the incident. If Claude Code half-loaded and then died on auth, that matches it too.
How loud it got in India
Downdetector is a complaint counter, not a census of accounts. Even with that limit, the Friday spike is why this sat on Indian tech desks. Around 3:45 pm IST the tracker logged 2,895 reports about Claude, against a baseline of two. Business Today is explicit that these are user-submitted outage reports, not a confirmed count of affected customers. Moneycontrol calls the same 2,895-versus-two reading a significant jump, and says complaints from India were part of what made the spike visible.
A baseline of two means a normal Friday afternoon barely registers. Jumping to nearly three thousand reports in the same window Anthropic posted its status note is why people assumed a total blackout. The dashboard says otherwise. Major on the chat site, partial on Code and Console, API up. The complaint wave measures pain at the login screen. It does not measure every API customer who never noticed.
What Anthropic has and has not said
The 3:28 pm IST note is narrow. Anthropic said it was investigating an issue affecting users attempting to authenticate with Claude.ai and Claude Code. It said it would share more information as soon as it had an update. It did not name a faulty server, a bad deploy, a region, or a cyber incident. Nobody in either article has published a root cause for the login failure. Do not invent one from the Microsoft outage next door.
Moneycontrol also says the company confirmed it was investigating elevated errors across multiple services, including its developer platform. That matches Console sitting on platform.claude.com. There is still no line that says “fixed”, and no line that says how many accounts were locked. If you are reading this after the chat loads again, treat that as your own check, not as a sentence Anthropic had given these two newsrooms by Friday evening.
The Microsoft 365 fault is a separate incident
This is the part that got messy in group chats. Claude for Outlook and the Microsoft 365 connector were already sick before Friday’s login failure, and Anthropic has said that sickness belongs to Microsoft.
On 12 November at 17:52 UTC, Anthropic first reported elevated errors and slow responses. It later identified an upstream provider and said it was working with that provider while affected servers were restarted. In an update at 22:23 UTC the same day, the company said the connector disruption was caused by a Microsoft 365 service incident, and pointed users to Microsoft’s own status page. Moneycontrol says Anthropic’s 12 November update explicitly attributed the connector trouble to that Microsoft incident. Mail, calendar and search tools inside the connector kept failing for most requests. SharePoint and Teams chat showed problems during the initial investigation. Claude for Outlook saw mail and calendar actions fail for some users, and some people could not open the add-in. Business Today’s Friday evening copy says a complete resolution of that connector issue had not yet been announced.
Read those timestamps against the login note and they do not merge. The connector trouble is a 12 November Microsoft 365 incident. The authentication errors on Claude.ai, Claude Code and Console were acknowledged on 13 November at 3:28 pm IST. Moneycontrol’s article says Anthropic has not linked the Microsoft incident to the login failure. Business Today says the login disruption is separate. If Outlook’s Claude add-in is dead and the website also will not log you in, you are looking at two open items, not one outage with two symptoms.
What to do if you still need Claude today
Neither desk published a workaround list, so this is only the split already on the status page. Retrying the website password does not fix an authentication incident, and it can lock a healthy account if you keep guessing. The API being operational means a product that already talks to Claude through an API key was not the thing these reports say broke. It does not mean you should go hunting for someone else’s key, and it does not mean a half-loaded Claude Code session will push through.
For the Outlook add-in, the path Anthropic already gave is Microsoft’s service status, not another Claude login attempt. That connector failure is the Microsoft 365 incident. Refreshing claude.ai will not repair a mail action that Anthropic has already blamed on Microsoft’s outage.
Until a newer status note exists, the public record from Friday evening is short. Investigating since 3:28 pm IST. No estimated restoration time. Claude.ai major. Claude Code and Console partial. Cowork degraded. API up. Downdetector at 2,895 reports around 3:45 pm IST against a baseline of two. Microsoft connector still a different ticket from the day before.
What nobody has published
No cause for the auth errors. No region list. No India-only user count beyond those 2,895 reports. No statement that paid plans were spared or hit harder. No compensation, no credit, no promise that chats typed during the failure were saved. Cowork’s “degraded” label is not explained. Console’s partial outage is not broken into which buttons failed. If a Telegram forward adds any of that, it is ahead of both newsrooms and ahead of Anthropic’s own note.
The line to drop in the work chat
Friday, 13 November 2026, 3:28 pm IST: Anthropic said it was investigating login failures on Claude.ai and Claude Code, with no restore time. Claude.ai was a major outage, Console and Claude Code were partial, Cowork was degraded, and the API stayed up. Downdetector showed 2,895 Claude reports around 3:45 pm IST, against a usual two. The dead Outlook connector is the earlier Microsoft 365 incident from 12 November, and Anthropic has not tied that to the login outage.
Comments (0)
Be the first to comment!