🔍 Read the full analysis: AI Chats And Zero Retention: Notion’s Account Of The Fireworks Breach on ThorstenMeyerAI.com
Get smart everyday buys delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Notion says data accessed in the Fireworks AI breach included Notion AI chat transcripts used for its Insights feature, affecting fewer than 1% of workspaces. The company says Fireworks had contractually committed to zero data retention, but Notion learned during its investigation that data from the endpoints it used had been retained. The companies have not explained which system stored it or for how long.
Notion says a dataset accessed in the Fireworks AI breach included chat transcripts from its AI product, and that data had been retained by Fireworks despite a contractual zero-retention commitment. Notion estimates that fewer than 1% of workspaces were affected and says it revoked exposed Notion API keys found in affected user messages on November 11.
Notion’s updated advisory says the exposed dataset contained Notion AI chat transcripts used for its Insights feature, which gives workspace administrators information about team use of AI. Notion says the transcripts were unattributable for most affected workspaces: they were not labeled with the person who wrote them or the workspace they came from, and did not include tool-call results. That does not mean the transcript content was harmless. The company defines a transcript as any input to or output from the model written to the chat, which could include sensitive details entered by users.
Notion says it used Fireworks for Insights classification jobs and as an inference provider for open-weight models in its product. It says the affected data came from the Insights work. Notion also says AI Meeting Notes, including audio, transcripts and summaries, were not affected unless that material was used as input in a Notion AI chat transcript. The company states that no other Notion features were affected.
Notion’s notice says the company disconnected its systems from Fireworks on November 3, deleted data from the Fireworks system and stopped communications with the provider. It also says it halted training runs. On November 11, Notion revoked exposed API keys found in affected user messages. The company advises affected users to treat the data as exposed, review it through its Security incident review process and rotate passwords, API keys or other credentials that may have been pasted into Notion AI during the incident.
Notion’s Fireworks breach and zero retention
Notion says retained Fireworks data included some Notion AI chat transcripts. The incident raises questions about what zero retention covers and how businesses protect sensitive AI inputs.
What data was involved
Notion says the dataset included chat transcripts used for its Insights feature.
AI chat content
Transcripts could contain anything users typed or received in chat, including sensitive details.
Limited labels
Most affected transcripts lacked the writer’s identity and workspace label. That does not make their content harmless.
What zero retention covered
Notion says its commitment covered live requests sent to generate responses, while stored data used for features such as Insights fell outside that scope.
- Zero retention for live model requests
- Fireworks had contractually committed to zero retention for Notion’s platform use
- Data from endpoints Notion used had been retained
- The storage system and retention period remain undisclosed
Notion’s response
The company says it cut off Fireworks access and took steps to address affected data and credentials.
Notion says it disconnected from Fireworks on November 3, deleted data from its system and halted training runs.
On November 11, Notion revoked exposed API keys found in affected user messages.
Affected users should review the material and rotate credentials they may have entered into Notion AI.
Why the Retention Failure Matters
The disclosure matters because it complicates the meaning of zero data retention for customers using AI features. Notion says its ZDR commitments cover live requests sent to model vendors to generate responses. It says they are not intended to cover features that require stored data, including chat history and Insights. For Insights, Notion says it sent stored requests and responses from Notion AI use to Fireworks for analysis.
Notion says Fireworks gave it a contractual commitment to zero data retention for its use of the platform, but that its investigation found data had been retained from the endpoints it used. The incident therefore raises a practical question for businesses evaluating AI vendors: which data is retained for product features, which systems process it, and whether contractual limits were followed. The published information does not establish that every Fireworks customer or every Notion AI user was affected.
For organizations whose workspaces were identified by Notion, the immediate concern is whether transcripts contain confidential material or credentials. Notion’s recommendation to review affected data reflects that transcript content can include what users typed or received, even if records do not identify the author or workspace.
Products Worth Considering
Canon SELPHY CP1500 Wireless Compact Photo Printer with RP-108 Ink and Paper Set, Black, 108 Sheets
Canon IVY 2 Mini Photo Printer Bundle, Blush Pink
Vari Ergo Electric Standing Desk, 54×26, Light Wood/White
As an affiliate, we earn on qualifying purchases.
How Notion Describes Its AI Data
Notion’s revised help page distinguishes between live model requests and stored data used for features such as chat history and Insights. The company says its zero-retention commitments apply to requests sent to third-party model vendors to generate Notion AI responses, while stored data used to deliver other features falls outside that stated scope.
Notion’s latest account follows Fireworks’ breach notice. Fireworks’ bulletin, according to the supplied source material, was last updated November 9 and said attackers accessed data including datasets for certain customers. Fireworks also said it had not detected an impact to Inference, Training SDK and Nexus, which it described as zero-retention by default. Notion has not identified which Fireworks system retained its data, and the available statements do not explain how the two companies’ descriptions relate.
Notion’s published timeline says Fireworks first notified it on November 2 that a small set of data had been accessed by malicious actors. Notion initially understood the data to be synthetic testing data. The company says it disconnected from Fireworks the next morning, before its later investigation identified the AI chat transcripts in the dataset.
“Unfortunately, during our investigation we learned that data was retained from the Fireworks endpoints that we used.”
— Notion, in its security advisory
What the Breach Notices Leave Open
The companies have not disclosed which Fireworks system retained Notion’s data, how long the data remained there, or how many individual users were affected. Notion gives a workspace estimate—fewer than 1%—but the notice does not provide a user count. It is also unclear whether other Fireworks customers experienced the same retention issue.
The notices do not reconcile Notion’s finding with Fireworks’ statement that it detected no impact to several services described as zero-retention by default. Neither company’s published account, as described in the source material, identifies the relevant endpoint or explains why data was stored. Notion’s characterization of transcripts as unattributable for most affected workspaces also does not establish that the content itself lacked sensitive information.
Vendor Audit and User Reviews
Notion says it is auditing critical vendors, including inference providers and data processors, to check their data-minimization practices and zero-retention policies. It has not published a completion date or said whether the review will result in changes to its vendor controls.
Users should follow Notion’s instructions if the company has contacted them or their workspace owner: treat affected data as exposed, review the material through the Security incident review process and rotate credentials entered into Notion AI during the incident. The next key developments would be further detail from Notion or Fireworks about the storage system, retention period, scope of affected accounts and any broader customer impact.
Key Questions
What Notion data was included in the Fireworks breach dataset?
Notion says it included Notion AI chat transcripts used for the Insights feature. The company says most affected workspaces’ transcripts were not labeled with the writer or workspace, but the content could still contain sensitive information.
How many Notion workspaces were affected?
Notion estimates that fewer than 1% of workspaces were affected. It has not published a precise number of workspaces or individual users in the source material.
What does Notion advise affected users to do?
Notion advises affected users to treat the data as exposed, review it through its Security incident review process and rotate passwords, API keys or other credentials entered into Notion AI during the incident.
Did the breach affect Notion AI Meeting Notes?
Notion says AI Meeting Notes—including audio, transcripts and summaries—were not affected unless that content was used as input in a Notion AI chat transcript. It says no other Notion features were affected.
How did data remain stored despite the zero-retention commitment?
That has not been explained. Notion says Fireworks had contractually committed to zero data retention but that its investigation found data retained from the endpoints it used. The companies have not identified the system involved or the length of retention.
Source: ThorstenMeyerAI.com
