Claude isn’t working. Since this morning, Friday 13 November, anyone trying to log into Anthropic‘s artificial intelligence assistant has been locked out. The company confirmed the fault at 10:58 am: it is investigating an access problem but has given neither a cause nor a timeline for restoring service. This comes as Outlook and other Microsoft services have been malfunctioning since yesterday.
Claude is one of ChatGPT’s main rivals. Shortly after 11 am, its official status page was still showing a “major outage” for the site used to access the chatbot. Claude Code, the tool used by developers, and the developer console have also been working intermittently. Cowork has been slowed down. The connection used by companies that integrate Claude into their own software, however, is holding up.
Users raised the alarm before the official confirmation. Downdetector, the site that collects service-disruption reports, logged problems starting at 10:51 am — seven minutes earlier. The outage knows no borders: it is also affecting Claude users in Italy and San Marino.
The timing is dreadful. Since late yesterday afternoon, Microsoft has been grappling with a global blackout affecting Outlook, Teams and other services, which had drawn more than 130,000 reports on Downdetector and had still not been resolved by this morning.
Microsoft has, for now, ruled out a cyberattack. Because of that blackout, Claude has been unable to read Outlook mail and calendars since last night — a fact confirmed by Anthropic itself.
As for today’s outage, no connection between the two has been established. Until proven otherwise, these appear to be two separate faults that simply happened at the same time.
This is not the first time something like this has happened. The monitoring service StatusPal counts 13 incidents over the past 30 days, not including this morning’s. On 29 September, another login outage generated more than 12,000 reports.
For now, all that’s left to do is wait: Anthropic has promised updates on its status page — the same page that, over the last 90 days, reports the service as having been up and running 99.75% of the time. The remaining 0.25 percent chose a Friday the 13th to make itself known.
