Fireworks AI updated its October 2026 security incident bulletin (last updated 9 November 2026). It says it notified law enforcement and engaged Palo Alto Networks Unit 42 for forensic analysis; when it published its first bulletin on 9 October 2026 only Fireworks-stored secrets (for example those used in evaluation jobs) were known to have been accessed, but it has since found impact to account and billing information and, for a limited set of accounts, to secrets, datasets and incomplete models.

Fireworks says the evidence indicates links to a financially motivated cybercriminal group that sells tokens from unauthorised access to large language models to illicit routing gateways ('LLMjacking').

It says a third-party cloud provider hosting Fireworks R&D machines had a security incident on 23 October, and on 24 October the unauthorised party used it to reach Fireworks development machines and run an unauthorised workload that caused more customer data impact; Fireworks detected it the same day and says it has not detected further successful unauthorised access to the customer platform as of 25 October. Data involved: resource metadata (identifiers, names, descriptions, creation times); billing information such as billing and shipping addresses, invoice and payment history, account balance and credit usage (payment card numbers are not stored and were not affected); and, for certain customers, a subset of incomplete model weights and datasets.

Fireworks says it found no impact to Inference, the Training SDK or Nexus, which are zero data retention by default; it is notifying affected customers, has revoked and replaced credentials, isolated affected systems, and disabled self-serve dataset uploads and downloads for Batch Inference and dataset-based training while it investigates. (source: Fireworks AI)

Updated (10 Nov 2026): Notion said in a help-centre notice that its investigation found fewer than 1% of Notion workspaces may have been affected by the Fireworks incident and that some unattributable Notion AI chat sessions, used for its Insights feature, may have been exposed; it says the chats were not labelled with who wrote them or which workspace they came from, contained no tool call results, and no other workspace data is affected.

Notion says Fireworks notified it on 6 November that malicious actors may have accessed potential customer data and confirmed the new dataset on 8 November; Notion has disabled Insights, no longer serves open-weight models through Fireworks, and has contacted owners of affected workspaces. (source: Notion)

Updated (12 Nov 2026): Notion's security bulletin (first published 9 November, updated 10 November 2026) adds that Fireworks first notified it on the evening of 2 November that a small set of data had been accessed; Notion says that dataset held Notion AI synthetic testing data and no Notion AI customer data, and that by the morning of 3 November it had disconnected its systems from Fireworks, deleted all data in the Fireworks system and stopped all communication with Fireworks. Notion recommends affected customers treat the data as if it were exposed and rotate any passwords, API keys or other credentials pasted into Notion AI during the incident timeframe; its help-centre notice says workspace owners can review affected data under Security incident review in a desktop browser. (source: Notion)