Service eligibility and regional restrictions
PuppyIP serves only compliant overseas businesses and their authorized personnel. Proxy services are not available in mainland China. The service may only be used for lawful business activities outside mainland China. Use of this service within mainland China is prohibited.
Hosting a proxy IP or server overseas does not change these restrictions. The service must not be provided to end users in mainland China through relaying, forwarding, sharing or resale. Before use, read the Terms of Service.
Key Takeaways
- OpenAI is investigating Microsoft 365 connector errors: new connections may be broken and tools may not work as expected.
- The first official update was displayed at 18:29 UTC on Nov 12, 2026, or 02:29 on Nov 13 in Shanghai. A check at 18:43 UTC still showed Investigating.
- The status report does not list affected Microsoft 365 services, accounts, tenants, plans or regions, and gives no cause, recovery ETA or verified official mitigation.
- Separate new-connection failures from errors in existing tool calls, keep redacted error details and check the latest official status. These are editorial process suggestions, not a verified fix.
An official incident is under investigation
OpenAI's status record lists incident 01M78GAT3AK9P9W8Z230KHFVE8, titled “Elevated errors affecting the Microsoft 365 connector.” Its status is Investigating, and the status API assigns the incident a Minor impact level. OpenAI says Microsoft connectors are having an issue: new connections may be broken, and tools may not work as expected.
The official record shows the incident was created at 18:29:14 UTC on Nov 12, 2026, with its first update recorded one second later. In Shanghai, those times were 02:29:14 and 02:29:15 on Nov 13. A fresh check at 18:43 UTC (02:43 in Shanghai) still showed Investigating; check the official incident page for subsequent updates.
What is confirmed, and what is not
The confirmed report concerns possible failures when creating new Microsoft connector connections and tools not working as expected. The status page does not name specific Microsoft 365 services, tenants, accounts, plans, models or regions. It does not say that every account or Microsoft 365 as a whole is unavailable.
Minor is the status system's aggregate impact label; it does not establish whether an individual account is affected. As of the 18:43 UTC check, OpenAI had provided no root cause, estimated recovery time or verified official workaround. Investigating does not mean the issue has been resolved.
Separate new connections from existing tool calls
First identify whether the error occurred while creating a new connection or during a tool call on an existing connection. Keep the time, client and error category, with sensitive details removed. Do not record access tokens, passwords, complete tenant files or other sensitive request content. This is editorial triage advice; OpenAI has not said that collecting these details will fix the incident.
Check the latest official incident page for its update time and current status, since the investigation can change. If the work can wait, consider deferring unnecessary repeated authorizations and automated retry loops so repeated requests are not mistaken for a repair. This is process advice, not an OpenAI-published workaround.
If an alternative way to access the work is already approved by your organization and available, you can use it. Do not assume that changing account permissions, region, plan or proxy settings will repair this connector incident; the current official sources provide no evidence for that.
Keep the incident separate from local evidence
OpenAI describes an issue with Microsoft connectors, but does not say that every Microsoft 365 service is affected. If the problem appears only for one account, workspace, client or existing authorization, the incident alone cannot establish the cause of that case. Keep local evidence separate from the scope stated in the announcement.
For local triage, compare the failure time, whether it was a first connection or a later tool call, the client involved and the organization's authorization state. Share redacted errors only through a support channel approved by your organization. This is editorial advice; no real-user test or local reproduction data is available here, and it cannot replace confirmation from the service provider.
Use the latest status to decide when to retry
Before retrying a new connection or tool call, check the latest update on the official incident page; this saved snapshot is not a current-status guarantee. If the incident is marked resolved but errors continue, investigate account, client, organizational authorization or local connection evidence separately. The available sources give no promised recovery time.
Sources
Frequently Asked Questions
Will a failed request complete automatically after the incident is resolved?
The announcement does not say that failed requests will be retried or completed automatically. As process advice, check whether the original task completed before deciding to retry through your organization's approved workflow. Service recovery does not establish that a particular task completed.
Does the incident announcement promise a refund or service credit?
As of this check, the incident record announces no fee adjustment, refund or compensation arrangement and gives no related eligibility criteria. The incident alone cannot establish whether an account qualifies under any separate existing terms.