By 
SAN FRANCISCO, United States, November 12, 2026 — Productivity software company Notion has disclosed that a security incident at its third-party artificial intelligence infrastructure provider, Fireworks AI, resulted in the exposure of certain Notion AI chat transcripts associated with a small number of workspaces. The company said the incident affected fewer than 1% of Notion workspaces and that it had contacted customers it identified as affected.
The disclosure demonstrates how security incidents involving external AI providers can create data-protection risks for companies that rely on third-party infrastructure to deliver product features.
How the Incident Developed
Notion said Fireworks notified it on November 2 that malicious actors had accessed a limited dataset. The initial investigation found that the information involved synthetic testing data rather than customer conversations. Notion said it disconnected its systems from Fireworks, deleted its data from the provider’s environment and stopped using the service for the relevant functions by November 3.
The situation changed after Fireworks notified Notion on November 6 that another dataset potentially containing customer information may have been accessed. Fireworks confirmed the new dataset to Notion on November 8. Notion’s subsequent review found that it included AI chat transcripts used for the platform’s Insights feature.
What Information Was Involved
Notion said that, for the vast majority of affected workspaces, the transcripts could not be attributed to a specific user or workspace. It also said the exposed transcripts did not include tool-call results and that no other Notion features were affected.
The company has advised affected customers to treat the relevant data as potentially exposed. It recommends rotating passwords, API keys or other credentials if sensitive information was entered into Notion AI during the incident period. Notion said it would contact customers whose workspaces it identified as affected and provide instructions for reviewing the relevant information.
Third-Party AI Security Comes Under Scrutiny
The incident adds to the growing challenge of securing AI applications that depend on external model providers, inference platforms and data-processing services. Even when a software company’s own systems are not breached, information processed by a vendor can create a separate exposure pathway.
For organisations adopting AI tools, the case reinforces the importance of understanding what information is sent to external providers, how long it is retained, how vendors handle security incidents and whether customer data can be traced to specific users.
Notion said its investigation and monitoring were continuing. The company has not reported that its wider platform or other product features were compromised.
Comments