Microsoft outage disrupts Copilot and breaks AI connections to Teams and Outlook

A major Microsoft outage caused disruptions across the globe on Friday, taking down core workplace applications and breaking automated connections used by external artificial intelligence platforms.

The incident began on Thursday evening Australian Eastern Daylight Time and lasted for roughly 18 hours before services started to return to normal. The interruption affected Exchange Online, Outlook, Teams and Copilot, locking out workers across financial firms, government agencies and schools.

Beyond standard email and chat, the technical failure spilled over into software integrations. OpenAI and Anthropic reported problems with their Microsoft 365 connectors, meaning AI tools that pull data from or take action inside corporate mailboxes and chat channels could not function as expected.

Technical cause rather than a cyber attack

Microsoft ruled out malicious activity as the reason for the downtime. Charles Lamanna, Microsoft executive vice president for Copilot, Agents and Platform, said on social media that there was no indication the problem was caused by a cyber attack or by autonomous AI agents.

Instead, the company pointed to excessive infrastructure demand in multiple regions. Engineers restarted affected back-end servers and rerouted network traffic to ease the pressure, though the recovery took longer than first projected. Microsoft confirmed it had resolved the core issue and was bringing systems back in phases to maintain network stability.

What this means for Australian businesses

Many small and medium companies in Australia rely on Microsoft 365 to handle everyday communications. When the central platform stalls, linked AI assistants fail along with it.

• Map third-party AI dependencies: Review every workflow where an AI model reads your team's emails, updates calendars or answers internal questions via Teams. Document what happens to those operations when access to Microsoft drops.
• Maintain direct fallbacks: Avoid routing all business automation through a single software suite. Staff should know how to access raw document files or alternative communication channels if automated connectors stop working.
• Check automated task queues after recovery: When connectors reconnect after an extended downtime, check that automated triggers did not misfire, drop draft customer replies, or run duplicate operations while processing backed-up queues.