🔍 Read the full analysis: Notion’s AI Chat Transcript Exposure: What The Fireworks Breach Shows on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
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 were affected and that it revoked exposed Notion API keys found in affected messages on November 11. The cause, retention period and scope beyond Notion remain unclear.
Notion says Notion AI chat transcripts were included in a dataset accessed during the Fireworks AI breach, and that the data had been retained despite Fireworks’ contractual zero-data-retention commitment. Notion says fewer than 1% of workspaces were affected and that, on November 11, it revoked exposed Notion API keys found in affected user messages.
According to Notion’s updated security advisory and Help Center notice, the transcripts were associated with its Insights feature, which gives workspace administrators information about how teams use AI. Notion said the dataset Fireworks provided to it on November 8 included chat transcripts from affected workspaces. For most affected workspaces, the transcripts were “unattributable”: Notion says they were not labeled with the author or workspace and did not include tool-call results.
That description does not mean the transcript content was necessarily harmless. Notion defines a transcript as input to or output from the model written in the chat, and advises affected customers to review the data because prompts or responses may contain sensitive information. Notion said AI Meeting Notes—including audio, transcripts and summaries—were not affected unless they had been used as input in a Notion AI chat transcript. It said no other Notion features were affected.
Notion’s notice says that, after learning data had been retained from the Fireworks endpoints it used, the company disconnected its systems from Fireworks on November 3, deleted data in the Fireworks system and halted training runs. On November 11, Notion also revoked exposed Notion 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 entered into Notion AI during the incident.
Notion AI Chat Transcripts Exposed in Fireworks Breach
Notion says a dataset accessed in the Fireworks AI breach included transcripts used for its Insights feature. The incident shows why businesses need to check how vendor data-retention commitments work in practice.
What Was in the Dataset
The transcripts were associated with Insights, which helps workspace administrators understand how teams use AI.
Chat inputs and outputs
Prompts or responses may contain sensitive information, Notion says.
Limited attribution
Most affected transcripts lacked author and workspace labels and tool-call results.
Other features
Notion says AI Meeting Notes and other features were not affected, unless meeting material was used in a Notion AI chat.
What Notion Did
The company says it acted after learning data had been retained from Fireworks endpoints it used.
Notion disconnected its systems from Fireworks on November 3.
It says it deleted data in the Fireworks system and halted training runs.
On November 11, Notion revoked exposed API keys found in affected messages.
The Retention Gap
Contractual promises and the handling of stored data may not cover the same workflows.
- Fireworks had a contractual zero-data-retention commitment.
- Some features, including chat history and Insights, involve stored requests and responses.
- Which Fireworks system retained the dataset, why, and for how long.
- Whether other Fireworks customers faced the same retention issue.
What Affected Customers Can Do
Notion advises customers who receive a notice to treat relevant data as exposed.
- Review affected data through Notion’s Security incident review process.
- Rotate passwords, API keys, or other credentials entered into Notion AI during the incident.
- Check whether transcripts included sensitive business or customer information.
The Gap in Zero-Retention Protections
The disclosure matters because businesses may rely on zero data retention commitments when deciding whether to send sensitive material to an AI service. Notion says Fireworks had made a contractual commitment not to retain data for Notion’s use of its platform, but that its investigation found data had been retained from the endpoints involved.
Notion also describes a distinction in its own approach: zero-retention protections cover live requests sent to model vendors to generate responses, while some features, including chat history and Insights, involve stored requests and responses. That distinction can affect what information is sent to a vendor for additional processing and what data may remain available for a feature. The incident does not establish that every AI interaction or every Notion customer was exposed, but it shows why the scope and operation of vendor commitments matter alongside their contractual wording.
For affected organizations, the immediate concern is practical: a chat transcript could contain customer details, contract information or credentials entered by a user. Notion has urged a review of affected data and rotation of credentials found in messages. The company has not said that all such sensitive information was present or misused; the potential exposure depends on what users entered.
Products Worth Considering

Samsung 32-Inch Odyssey G5 QHD 180Hz Gaming Monitor

Dell UltraSharp U4025QW 40 5K2K WUHD Curved LED Monitor (21:9)

Apple iPad 11-inch (A16), 128GB, Wi-Fi, Silver
As an affiliate, we earn on qualifying purchases.
How Notion Used Fireworks
Notion says it used Fireworks for two purposes: classification jobs for Insights and as an inference provider for open-weight models in its product. The transcripts at issue came from the Insights work, rather than from all uses of Notion AI, according to the company’s notice.
Notion’s published timeline says Fireworks first told 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. On November 3, it says, it disconnected from Fireworks, deleted data held in the Fireworks system and halted training runs. Fireworks later told Notion that additional data may have been accessed; Notion’s subsequent updates added the transcript disclosure and the API-key revocations. Fireworks’ own bulletin, as described in the source material, was last updated November 9 and said certain customer datasets had been accessed.
The two companies’ published descriptions leave an unresolved technical question. Fireworks said it had not detected an impact to Inference, Training SDK and Nexus, which it describes as zero-retention by default. Notion says data was retained from the Fireworks endpoints it used. Neither company has identified which Fireworks system held Notion’s data or explained how the statements fit together.
“Unfortunately, during our investigation we learned that data was retained from the Fireworks endpoints that we used.”
— Notion
Scope and Retention Details Still Unknown
Notion has not disclosed which Fireworks system retained the data, why it was retained, or how long it remained there. The companies have not provided a complete count of affected people or a detailed breakdown of the affected workspaces beyond Notion’s estimate that they represent fewer than 1% of its workspaces.
It is also not clear from the published notices whether other Fireworks customers experienced the same retention problem. The available information does not establish whether attackers accessed or used particular sensitive items within the transcripts, or whether any credentials were used after exposure. Notion’s account and Fireworks’ description of systems with zero retention by default do not resolve where the retained dataset was stored.
Notion says a workspace has not been identified as affected if neither its owner nor users have received notice from the company. That is Notion’s stated notification guidance; the public material provided does not include a full list of affected customers or independently verified details of the investigation.
Vendor Audit and Customer Reviews
Notion says it is auditing critical vendors, including inference providers and data processors, to check their data-minimization practices and zero-retention policies. The company has not published a completion date or specific findings from that review. Further updates could clarify what Fireworks endpoint retained the data and how the retention occurred.
For customers who receive a notice, Notion’s guidance is to treat relevant data as exposed, examine it through the company’s Security incident review process and rotate any passwords, API keys or other credentials pasted into Notion AI during the incident. Organizations can also assess whether their affected transcripts contained information requiring additional internal response. Notion has not said that every affected transcript contained sensitive material.
The next public milestones are any further updates from Notion or Fireworks on the investigation, the affected data and the vendor audit. Until those details are published, the reported workspace percentage and the companies’ descriptions should not be read as answers to the open questions about retention, access or the wider Fireworks customer base.
Key Questions
What information was involved in the Notion AI exposure?
Notion says a dataset accessed in the Fireworks breach included Notion AI chat transcripts used for Insights. A transcript can include prompts and model responses written in the chat; the company says most affected transcripts were not labeled with an author or workspace.
How many Notion workspaces were affected?
Notion says fewer than 1% of workspaces were affected. The available notice does not provide a precise number of workspaces or people.
Were AI Meeting Notes affected?
Notion says AI Meeting Notes—including audio, transcripts and summaries—were not affected unless they were used as input in a Notion AI chat transcript. It also says no other Notion features were affected.
What should affected users 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.
Why was data retained despite a zero-retention commitment?
That has not been explained in the published notices. Notion says Fireworks had made a contractual zero-retention commitment but that its investigation found data had been retained from the endpoints it used. Neither company has identified the system involved or said how long the data was kept.
Source: ThorstenMeyerAI.com
