Notion Discloses AI Chat Exposure Via Fireworks
Notion says transcripts tied to fewer than 1% of customer workspaces were exposed in a security incident at its AI infrastructure provider, Fireworks AI.
Updated on Nov. 10, 2026 in Artificial Intelligence
Live Poll
Do you trust AI tools to keep your chat conversations private?
Notion confirmed that Notion AI chat transcripts linked to fewer than 1% of its customer workspaces were exposed after attackers accessed a small dataset held by Fireworks AI. Notion says its own systems were not breached and that no other product features were affected.
Why it matters
The incident shows how third-party AI providers can expose customer conversations even when a platform's own systems remain intact. Users of AI features embedded in workplace software face data risks that depend on the security of every vendor in the chain.
The exposed material consisted of Notion AI chat transcripts that Insights used, and Fireworks had been running classification jobs and open-weight models for Notion. The total number of exposed transcripts was not disclosed.
The players
Notion
Notion is a workplace productivity and collaboration platform whose AI features, including Notion AI and Insights, process user content.
Fireworks AI
Fireworks AI is an inference platform company that hosts and serves AI models for other software businesses.
The details
Fireworks first alerted Notion on November 2, 2026, and Notion's initial review found only synthetic testing data. After Fireworks warned on November 6 that customer data may have been accessed and confirmed the new dataset on November 8, Notion disconnected from Fireworks, deleted the data held there, and disabled Insights and open-weight model serving through Fireworks.
Timeline
November 2, 2026: Fireworks first alerts Notion to the incident.
November 3, 2026: Notion disconnects from Fireworks and deletes data held there.
November 6, 2026: Fireworks warns attackers may have accessed customer data.
November 8, 2026: Fireworks confirms the new dataset containing customer transcripts.
Roadmap
Many software platforms now route AI features through outside inference providers rather than running models in-house, which spreads customer data across more companies. This incident shows that outsourcing AI processing extends a platform's security perimeter to each of its vendors.
Users whose workspaces used Notion AI features should review what chat content those features processed. The incident underscores that AI tools embedded in everyday software can route private conversations through vendors outside the platform's direct control.
The takeaway
Users of AI features in workplace software should assume that conversations may pass through several vendors and consider what sensitive content they enter. Vendor security is part of the risk profile of any AI tool.
What happens next
Notion says it will contact any additional customers if its ongoing investigation finds further exposure.
Further reading
For more on how AI tools handle user data, see Artificial Intelligence.
Live Poll
Do you trust AI tools to keep your chat conversations private?
