AI service incidents · Daily incidents

AI incidents: Microsoft 365 connector recovery, Claude errors and OpenAI billing — 14 November 2026

Microsoft 365, Claude and ChatGPT posted connector recovery updates. Separate Claude access, OpenAI billing and GitHub Copilot incidents affected users.

Publisher disclosure: MUVO offers competing music, remix, and video tools. This report summarizes public sources and provider documentation; it is not an independent product test or a claim that MUVO is free of similar issues. Read our reporting method.

Microsoft 365 and its AI connectors posted recovery updates

Microsoft marked its Exchange-dependent Microsoft 365 incident restored in a tenant update at 07:17 Istanbul on 14 November. It said restoration was continuing for remaining customers and regions. Microsoft traced the issue to an earlier network capacity upgrade and said a full post-incident report would follow. Microsoft 365 Copilot Chat and features that access work mail, calendars and files were among the affected services. Microsoft tenant notice reproduced by Penn IT.

Anthropic reported that its Microsoft 365 connector and Claude for Outlook returned errors for many requests from 19:03 on 12 November to 19:40 on 13 November, an interval of 24 hours 37 minutes. It posted resolution at 00:49 on 14 November and warned of occasional residual delays. OpenAI posted signs of recovery at 19:20 on 13 November and marked its connector incident resolved at 03:18 on 14 November. The two notices cover their respective integrations. Anthropic notice · OpenAI notice.

Update, 16 November: Microsoft confirmed publicly on 14 November that service had been restored for the remaining affected customers and regions. The confirmation gave no separate time for the last recovery. Microsoft statement reproduced by Digital Trends · Associated Press report.

We added the follow-up to the 13 November report, including Microsoft’s formal record times.

Claude resolved separate authentication, service and purchase incidents

Incident First notice (Istanbul) Resolution or status at cutoff
Authentication and access 13 Nov 12:58 Resolved 13 Nov 14:58
Errors across Claude, API and Claude Code 13 Nov 16:09 Resolved 13 Nov 18:38
Purchases and plan changes 14 Nov 05:54 Resolved 14 Nov 07:23
Sign in with Apple on iOS 14 Nov 10:11 Investigating, mainly new sign-ups

For the authentication incident, Anthropic reported a fix at 14:05 on 13 November before closing the case. In the separate afternoon incident, Anthropic reported errors across Claude web and mobile, the API and Claude Code. It also noted disruptions to connectors and delays to some scheduled jobs. At 18:33 Istanbul it said core error rates had returned to normal while several features continued recovering. The notices do not identify Microsoft 365 as the cause of either incident. Claude incident history.

Anthropic reported a 05:34–07:03 Istanbul impact interval for web purchases and plan changes on 14 November and said delayed payments and plan changes were applied. For the separate iOS sign-up issue, Anthropic suggested Google or email for new accounts. Purchases notice · iOS notice.

OpenAI billing recovered while cost reporting remained delayed

OpenAI posted monitoring at 07:11 Istanbul and resolution at 08:27 on 14 November for errors affecting ChatGPT billing, including checkout, and Ads invoice retrieval. The notice does not give an impact start. Billing incident.

A separate delay in organization cost data was acknowledged at 19:10 on 13 November and moved to monitoring at 21:56. At 08:31 on 14 November, OpenAI still expected publishing delays and said ordinary API and ChatGPT usage were unaffected. Cost data incident.

GitHub Copilot recovered after errors on selected GPT models

GitHub reported elevated errors for GPT-6.1 Terra, GPT-5.6 Sol and GPT-6 Luna from 22:05 on 13 November to 00:27 on 14 November Istanbul. Its final notice attributes the degradation to an upstream model provider and says service returned after mitigation. GitHub suggested selecting another model or Auto during the incident. This notice covers GitHub Copilot. GitHub incident.

Fireworks and Bolt resolved separate component issues

Fireworks closed degradations on its hosted DeepSeek V4.1 Flash 1017 at 14:56, Kimi K3 Fast at 17:21, and GLM 5.3 Flash at 21:51, all on 13 November. The notices cover Fireworks inference. Separately, Bolt posted resolution for Prompting degradation at 19:09 the same day.

We also added Fireworks’ 10 November final closure to the original 4 November coverage. Fireworks closure notice.

Sources

Sources include the Microsoft tenant notice reproduced by Penn IT, Anthropic status archive and OpenAI incident history. All displayed times are Istanbul (UTC+3). First notice, monitoring and closure timestamps describe updates unless the provider specifies an impact window. Public reviews did not yield corroborated customer allegations for this edition. Some review platforms restricted access. Reporting methodology.

Update timeline