🔍 Read the full analysis: Notion AI Privacy: Fireworks Breach Reportedly Exposed Chat Transcripts on ThorstenMeyerAI.com
Get the little things that make your day 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 a dataset accessed in the Fireworks AI breach included Notion AI chat transcripts used for its Insights feature. The company says fewer than 1% of workspaces are affected and that it revoked exposed Notion API keys on November 11, but has not identified which Fireworks system retained the data or for how long.
Notion says attackers who accessed data at Fireworks AI obtained a dataset containing some Notion AI chat transcripts used for its Insights feature, and that the data had been retained despite Fireworks’ contractual zero-data-retention commitment. Notion says fewer than 1% of workspaces are affected and that it revoked exposed Notion API keys found in affected user messages on November 11.
In its updated security advisory and help-center notice, Notion said the dataset Fireworks provided to it on November 8 included transcripts from Notion AI chats used for Insights, a feature that helps workspace administrators understand how teams use AI. Notion says most affected transcripts were “unattributable”: they were not labeled with the author or workspace and did not include tool-call results.
That description does not mean the transcript contents were necessarily harmless. Notion defines a transcript as any input to or output from the model written in the chat. The company advises affected customers to review the data because a prompt or response may contain sensitive information, such as business details or credentials. Notion says AI Meeting Notes were not affected unless their audio, transcript or summary was used as input in a Notion AI chat, and says no other Notion features were affected.
Notion said on November 11 it revoked exposed Notion API keys that it found in affected user messages. It advises customers who receive an incident notice to treat the relevant data as exposed, review it under its Security incident review guidance, and rotate passwords, API keys or other credentials pasted into Notion AI during the incident.
Fireworks breach exposed Notion AI chat transcripts
Notion says a dataset accessed in the Fireworks AI breach included transcripts used for Insights. Prompts and responses may contain sensitive business information or credentials.
What the dataset contained
The transcripts came from Notion AI chats used for Insights, which helps administrators understand how teams use AI.
Inputs and outputs
Notion defines transcripts as anything written in the chat as input to or output from the model.
Mostly unattributable
Most affected transcripts lacked author and workspace labels and did not include tool-call results.
Limited feature impact
AI Meeting Notes were not affected unless their content was used in a Notion AI chat. Notion says no other features were affected.
What Notion has done
Notion says it disconnected its systems on November 3, deleted data in the Fireworks system, stopped communications, and halted training runs.
On November 11, Notion revoked exposed API keys it found in affected user messages.
Notion says it is reviewing vendors’ data-minimization practices, including zero-retention policies.
What remains unclear
The public notices leave key details about retention and scope unresolved.
- Fireworks retained data despite a contractual zero-retention commitment.
- Fewer than 1% of workspaces are affected.
- Which Fireworks system retained the data or how long it was stored.
- The exact number of affected workspaces, users, or records.
- Whether other Fireworks customers had the same issue.
What Transcript Exposure Means for Teams
The disclosure matters because chat transcripts can contain information users entered into an AI tool, not just generic usage statistics. Notion itself warns that prompts and responses may include sensitive data. Even when transcripts are not labeled with a person or workspace, their contents could still reveal business information or credentials. Notion’s warning to review affected data reflects that distinction.
The incident also raises a vendor-oversight issue for companies adopting AI services. Notion says it had a contractual zero-retention commitment from Fireworks, yet learned during its investigation that data from the endpoints it used had been retained. The disclosure does not establish why the retention occurred or whether the same issue affected other customers, but it highlights the gap that can emerge between a contractual promise and data handling in practice.
Notion says it is auditing critical vendors, including inference providers and data processors, to check their data-minimization practices. For affected organizations, the immediate concern is to identify what users put into chats and whether any credentials need to be changed.
Products Worth Considering
Hyperice Normatec Elite Wireless Compression Boots, Standard Size
Levoit EverestAir-P Smart Air Purifier for Large Rooms
Levoit Vital 100S-P Air Purifier
As an affiliate, we earn on qualifying purchases.
How Notion Used Fireworks
Notion says it used Fireworks for two kinds of work: classification jobs for Insights and inference for open-weight models inside its product. The transcript dataset described in the updated notice came from the Insights work. Notion says it disconnected its systems from Fireworks on November 3, deleted data in the Fireworks system, stopped communications between the systems and halted training runs.
Notion’s notice also explains that its zero-retention language applies to live requests sent to third-party model providers to generate responses. It says that commitment was not intended to cover features requiring stored data, including chat history and Insights. Separately, Notion says Fireworks had given it a contractual zero-retention commitment for the platform use involved here, but that its investigation found data had nevertheless been retained.
Fireworks’ own bulletin, last updated November 9 according to the supplied source, said it had not detected an impact to Inference, Training SDK and Nexus, which it described as zero-retention by default. Fireworks also listed datasets among the data accessed for certain customers. The published information does not explain how that account relates to Notion’s finding about retained data.
“Unfortunately, during our investigation we learned that data was retained from the Fireworks endpoints that we used.”
— Notion, in its updated incident FAQ
What Notion Has Not Identified
The public notices do not identify which Fireworks system retained Notion’s data, how long it was stored, or how many individuals or records were involved. Notion has given a workspace-level estimate — fewer than 1% — but the supplied information does not state a precise number of affected workspaces or users.
It is also unclear whether other Fireworks customers experienced the same retention issue. Fireworks’ bulletin and Notion’s account describe different aspects of the incident, but neither published notice, as presented in the source material, explains how the retained Notion data fits with Fireworks’ statement about its zero-retention-by-default services. The available facts do not establish the cause or extent of the discrepancy.
Notion says workspaces it has not contacted have not been identified as affected. That is the company’s current identification status, not independent confirmation that no further data or customers will be implicated as the investigation proceeds.
Vendor Audit and Customer Review
Notion says it is auditing critical vendors for data-minimization practices, including compliance with zero-retention policies. It has not provided a date for completing that work or published a full account of which Fireworks system retained the dataset. Further updates from Notion or Fireworks may clarify the system involved, retention period and scope.
Customers who have been notified should follow Notion’s incident-review guidance and check affected chats for credentials or other sensitive material. Notion recommends rotating passwords, API keys and other credentials that were entered during the incident. Workspaces that have not received a notice can monitor Notion’s security advisory for changes; the current notice says fewer than 1% of workspaces have been identified as affected.
Key Questions
What information does Notion say was exposed?
Notion says a dataset accessed in the Fireworks breach included Notion AI chat transcripts used for Insights. It says most affected transcripts were not labeled with a user or workspace, but warns their inputs or outputs could still contain sensitive information.
How many Notion workspaces are affected?
Notion says fewer than 1% of workspaces are affected. The published information supplied here does not give an exact number of workspaces or people.
Does Notion say AI Meeting Notes were affected?
Notion says AI Meeting Notes were not affected unless their audio, transcript or summary was used as an input in a Notion AI chat. It says no other Notion features were affected.
What should notified customers do?
Notion advises affected customers to treat the relevant data as exposed, review it under its Security incident review guidance, and rotate passwords, API keys or other credentials pasted into Notion AI during the incident.
What is still unknown about the Fireworks incident?
The notices do not specify which Fireworks system retained the data, how long it was kept, the precise number of people affected, or whether other Fireworks customers faced the same issue.
Source: ThorstenMeyerAI.com
