Fireworks AI breach exposed customer datasets and Notion AI chats

Fireworks AI widened its October breach disclosure on November 9, 2026 to cover billing data and, for a limited set of accounts, secrets, datasets and incomplete models. Notion says its AI chat transcripts were retained there.

Christopher Ort
By Christopher Ort · 3 min read
Share
A dimly lit server room with rows of rack-mounted equipment, emergency lighting casting long shadows across the floor
Image: AI-generated, i10X
On this page
  1. Quick Take
  2. In plain English
  3. What happened
  4. Who is affected
  5. What you should do
  6. What it means
  7. Sources

Fireworks AI updated its October 2026 security bulletin on November 9, 2026, expanding the scope of a breach that began when attackers compromised a third-party cloud provider and then moved laterally into Fireworks development machines. The company now says the incident exposed account and billing information, and for a limited set of accounts, secrets, datasets and incomplete models.

i10X Quick Take

Quick Take

  1. What happened: Attackers compromised a third-party cloud provider on October 23, 2026, then used that access to reach Fireworks AI development machines on October 24 and run unauthorized workloads.
  2. What was exposed: Account and billing information for all customers; secrets, datasets and incomplete model weights for a limited set of accounts.
  3. Notion impact: Fewer than 1% of Notion workspaces may be affected; Notion AI chat transcripts were retained on Fireworks endpoints despite a contractual zero data retention commitment.
  4. What to do: Rotate any secret stored on Fireworks and treat affected datasets as exposed; Notion users should check Security incident review in the sidebar and rotate any password or API key pasted into Notion AI during the affected period.
  5. Not affected: Fireworks Inference, Training SDK and Nexus, which the company says are zero data retention by default.

In plain English

Fireworks AI is a company that provides AI model hosting and development tools. In late October 2026, hackers broke into a cloud provider that Fireworks used for research and development, then jumped from there into Fireworks’ own systems. The attackers stole customer billing information and, for some customers, also grabbed API keys, datasets and partially trained AI models. Notion, which used Fireworks to power some of its AI features, discovered that Fireworks had kept copies of Notion AI chat transcripts even though their contract said Fireworks would not store any data.

What you should do: If you use Fireworks AI, you should change any passwords or API keys you stored there. If you use Notion, check the Security incident review section in your Notion sidebar to see if your workspace was affected, and change any passwords or API keys you typed into Notion AI between late October and early November 2026.

What happened

Fireworks AI first disclosed the incident on October 9, 2026, stating that it involved secrets used in evaluation jobs. On November 9, 2026, the company widened the disclosure to include account and billing information for all customers, and for a limited set of accounts, secrets, datasets and incomplete models.

A third-party cloud provider hosting Fireworks research and development machines had its own security incident on October 23, 2026. The next day, an unauthorized party used that access to reach Fireworks development machines and run an unauthorized workload. Fireworks ties the activity to a financially motivated group that resells LLM tokens obtained through unauthorized access, a pattern the company calls LLMJacking.

Notion published a separate notice the same week saying fewer than 1% of its workspaces may be affected. Notion states it held a contractual zero data retention commitment from Fireworks, then learned that data was retained on the Fireworks endpoints Notion used. Notion disconnected its systems from Fireworks as of November 3, 2026.

Who is affected

Fireworks customers whose accounts held secrets, datasets or incomplete model weights are in scope. Fireworks says affected customers are being notified directly with impact details.

Fireworks states that Inference, Training SDK and Nexus are zero data retention by default and that no impact to them has been detected.

Notion names its Insights feature, its open-weight model serving and its Learning and Early Access Program as the paths that sent data to Fireworks.

What you should do

Rotate any secret stored on Fireworks and treat affected datasets as exposed.

Notion tells workspace owners to open Security incident review in the Notion sidebar to see what was involved, and to rotate any password or API key pasted into Notion AI during the affected period.

What it means

The incident highlights supply chain risk in AI infrastructure. A breach at a third-party cloud provider cascaded into Fireworks AI’s development environment, then exposed customer data including secrets and datasets. The Notion disclosure adds a data retention dimension: despite a contractual commitment to zero data retention, Fireworks retained Notion AI chat transcripts, which became accessible to the attackers.

Fireworks has disabled self-serve dataset uploads and downloads and is conducting an investigation with Unit 42.

Sources

Christopher Ort

Christopher Ort

Christopher Ort handles SEO, AI agent workflows and PR & partnerships at i10x.ai, the all-in-one AI workspace built by i10X Global Pte. Ltd. in Singapore. He is a graduate of Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU). He is the author of “The Style Penalty”, a study published on SSRN that analyses 1,576 evaluation decisions by four frontier AI models to show how AI-written CVs fare in automated hiring screens; the research has been covered by heise online and t3n.

Editor's picks

  1. CVE-2026-126681: Prompt injection flaw in AWS Labs Timestream for InfluxDB MCP Server leaks API tokens
  2. CVE-2026-115563: OAuth credential leak in Model Context Protocol Python SDK
  3. US and Nine Allies Sign San Francisco Declaration to Protect AI Supply Chains
  4. Apple Launches Health AI Study With OpenEvidence for Up to 200,000 iPhone Users
  5. AMD Completes $8.2 Billion World Labs Acquisition, Fei-Fei Li Becomes Chief Scientist