Fireworks AI says its October breach reached customer billing data, secrets, datasets and incomplete model weights

Fireworks AI · Incident disclosure

Read the source (opens in a new tab)
What happened
AI model hosting company Fireworks AI updated its October 9, 2026 security bulletin on November 9, saying the breach reached more customer data than first known, including through an October 24 intrusion via a third-party cloud provider.
Key points
  • Fireworks says it found impact to account and billing information and, for a limited set of accounts, to secrets, datasets and incomplete models.
  • For certain customers, "a subset of incomplete model weights and datasets were accessed"; Fireworks says it does not store payment card numbers.
  • Fireworks links the attackers to a financially motivated group selling stolen access to large language models, known as "LLMJacking," and points to Google's November 2 report on it.
  • Fireworks says it detected the October 24 intrusion the same day and has disabled self-serve dataset uploads and downloads while it investigates.
Evidence
The company's own incident update, prepared with forensic help from Palo Alto Networks Unit 42. It does not name the group or say whether the attackers used AI.
Uncertain
Fireworks says its findings are "based on our investigation to date" and that the investigation is ongoing and may reveal more.

See it in the latest record