Less than 1% of Notion workspaces were impacted by a security breach at Fireworks AI, a global infrastructure platform used by Notion to power its AI features. While the breach potentially exposed some chat transcripts, Notion clarified that the leaked data was 'unattributable,' meaning the transcripts lacked labels identifying specific users, companies, or workspaces.

The Fireworks AI Infrastructure Breach

Notion utilized Fireworks AI as a primary infrastructure provider for two main functions: powering 'Insights' classification jobs to help customers track AI usage, and serving as a general inference provider for open-weight models. Following reports of a potential security incident, Notion proactively shut down all connections to Fireworks systems out of an abundance of caution.

On November 6, Fireworks notified Notion that malicious actors may have accessed customer data. A subsequent investigation confirmed a specific dataset containing Notion AI chat transcripts was included in the breach. In response, Notion halted all training runs and deleted all remaining data from the Fireworks system.

“Unattributable" means the chats weren't labeled with who wrote them or which workspace they came from, and the transcripts did not contain tool call results.”

— Notice about the Fireworks AI incident that may impact some Notion workspaces – Notion Help Center

Scope of Exposed Data

The security team determined that the exposed transcripts were 'unattributable.' This means the data did not contain information tying the chats back to a specific workspace, user, or company, and the transcripts did not include tool call results, such as Notion page content or external data pulled in by the AI.

Despite the lack of identifying metadata, Notion advises users to treat the data as if it were fully exposed. This precaution is intended to protect against any sensitive information, such as credentials or private notes, that a user might have typed directly into the chat interface.

Remediation and User Actions

Notion has provided specific tools for affected workspace owners to review the impact. Owners on Enterprise plans can access a CSV of the affected data, as well as an annotated CSV that includes metadata—such as the associated user—to help them run internal impact analyses. These annotations were not exposed during the breach.

For users on other plan types, workspace owners can share a link that allows members to review a CSV of their own specific data. Notion explicitly recommends that any passwords, API keys, or other credentials entered into Notion AI during the incident period be rotated immediately.

What Hacker News readers are saying

  • Alternative: One commenter argued that the incident highlights how a company's security is dependent on its third-party integrations, suggesting that businesses must prioritize risk management for external AI vendors. comment ↗
  • Skeptical: Several commenters expressed frustration that Notion required users to be on an Enterprise plan to identify which members were affected by the breach. comment ↗
  • Experience: One user shared their experience of having to manually identify affected members and share specific pages so that individuals could download their own data. comment ↗

Summarized opinions from the Hacker News thread, not verified facts. Read the full discussion ↗