Synopsis#synopsis

On March 2, 2026 at 23:30:24 UTC, we experienced an issue where the Zoom AI scribe was unable to join calls, rendering Zoom meeting transcription unavailable for all users. The issue persisted from approximately February 28 through March 5, 2026. Additionally, for a short period on March 5, 2026 conference bridge creation was not available.

Outside of a short window on March 5, Zoom conference bridges remained fully functional throughout the incident. Google Meet transcription was not impacted. Incident Management functionality in both the web UI and Slack was not impacted. Signals delivery was not impacted. Users were able to receive alerts and declare incidents with no issue.

Root Cause#root-cause

During an internal IT migration of FireHydrant Zoom accounts to Freshworks, the FireHydrant developer account that owned the Zoom Marketplace application was decommissioned. This broke the credentials used to connect to the Recall.ai transcription service, which powers the Zoom scribe feature. As a result, all attempts by recall bots to join Zoom meetings failed.

Timeline (UTC)#timeline-utc

• ~February 28, 2026 — All Zoom scribe bot join attempts begin returning zoom_invalid_signature errors.. The failure is not immediately detected because Zoom conference bridges continue to function normally — only the scribe/transcription integration is affected.
• March 2, 2026 at 23:30:24 UTC — The issue is detected internally via manual identification.
• March 2 – March 4, 2026 — Investigation traces the failure to the decommissioned developer account.
• March 5, 2026 at 14:29:17 UTC — Root cause formally identified and confirmed with Zoom support. Zoom backend engineering team confirms the application migration path is complete. A mitigation step was introduced to allow unauthenticated bots to join calls, restoring partial service. During the migration, Zoom credentials were refreshed which caused some customers to see runbook errors for Zoom steps failing to run.
• March 5, 2026 at 14:50:26 UTC — FireHydrant updates Zoom credentials and runbook steps begin working as expected.
• March 5, 2026 at 20:16:45 — Fix fully confirmed after monitoring period. All Zoom scribe functionality restored. Incident resolved.

Total time to resolve: 2 days, 20 hours, 46 minutes.

Customer Impact#customer-impact

Zoom transcription was completely unavailable for all users relying on the Zoom scribe feature.

Zoom conference bridges continued to function normally with a minor lapse in service. Google Meet transcription remained fully operational and was available as a workaround.

Users were able to declare and manage incidents with no issue on Slack and MS Teams.

All Signals notifications were delivered as expected with no delays.

Recovery:#recovery

• Two resolution paths are identified:
  • Primary path: Migrate the existing Zoom Marketplace application to the new Freshworks developer account. Escalated to Zoom's backend engineering support team.
  • Backup path: Create a new Zoom Marketplace application under the Freshworks account. Approval process initiated, but deprioritized due to Zoom's multi-week approval timeline and extensive back-and-forth requirements.
  • Customer-facing mitigations deployed: product banner directing users to the status page, CSMs notified and working with affected customers, Google Meet transcription confirmed as a functional workaround.

The primary resolution path was to migrate the existing Zoom Marketplace application to the Freshworks developer account and refresh Zoom SDK credentials. This effort was escalated to Zoom's backend engineering support team. Updated Zoom scribe settings were tested and the fix was confirmed on March 5, 2026.

All affected systems were returned to normal operations by March 5, 2026 at 20:16:45 UTC.

Corrective Measures#corrective-measures

We are applying the learnings from this incident to upcoming integration migrations, including a planned Slack integration migration. Specifically, we are implementing a more comprehensive service account and developer integration discovery process to ensure that developer tooling dependencies are identified and preserved during future IT migrations.

Next steps include:

1. Strengthening Observability - alerting for any failures around AI transcription (or any such critical operations) for proactively detecting the issues by systems. ( to reduce TTD)
2. Action plan for generating & associating AI transcriptions for the zoom conferences happened for any customer incidents during this FH incident window.
3. Following proper Change Management process for future integration migrations with detailed rollout & rollback plans

Commitment to Reliability#commitment-to-reliability

We understand that you rely on our services for your critical operations. We are committed to maintaining the highest levels of service reliability and transparency, and encourage you to subscribe to our Status Page for timely communication. The measures we're implementing will help ensure we continue to meet the standards you expect from us.