AI Stack Alert Medium TD-2026-0063 For CISO, AI Eng, AppSec AI-assisted
Notion Customer AI Chat Transcripts Exposed in Fireworks AI Breach
Notion is notifying a small set of customers that Notion AI chat transcripts used for its Insights feature were accessed in a breach at Fireworks AI. Fireworks AI was hacked twice in October, including a October 24 incident in which attackers ran an unauthorized workload that led to customer data compromise. Fewer than 1% of Notion workspaces were affected, and most transcripts could not be tied to a customer or workspace.
Why it matters
Users who pasted passwords, API keys, or other credentials into Notion AI may have exposed them to attackers. Organizations using third-party AI inference providers may have data exposed outside their own control.
Do this
- Identify whether any Notion AI workspaces in your organization were among those notified and review the retrieval instructions Notion provided.
- Rotate any credentials, passwords, or API keys that were pasted into Notion AI during the incident window.
- Treat Notion AI chat transcripts from the affected period as exposed and review them for sensitive content.
- Inventory third-party AI inference providers in use and confirm their breach notification and credential-handling commitments.
Signals
- NoCISA KEV
- 26/100Priority
- 70%Confidence
Evidence
“Fewer than 1% of Notion workspaces were affected, and most of the transcripts were unattributable to a customer or the workspace they came from.”
“We recommend treating this data as exposed. If you pasted sensitive information, passwords, API keys, or other credentials into Notion AI during the incident, rotate any exposed credentials,”
Sources
Affected / related
Notion, Fireworks AI, LLMJacking
Published 2026-11-12 11:08 UTC · updated 2026-11-12 17:05 UTC · AI-assisted summary, checked against quoted sources · How we score