---
title: Fireworks AI widens its October breach: billing data, and for some customers secrets, datasets and incomplete model weights; it ties the actor to LLMjacking | ctxsec
url: https://ctxsec.com/items/fireworks-ai-incident-update-llmjacking-billing-datasets-incomplete-model-weights
description: "Fireworks first disclosed on Oct 9 that internal credentials were used to read secrets in evaluation-job environment variables. Its Nov 9 update says the affected internal application also exposed resource metadata and billing information (addresses, invoices, balances, credit usage, but no card numbers), and that for a limited set of accounts secrets, datasets and incomplete model weights were accessed. A separate path followed: a third-party cloud provider hosting Fireworks R&D machines had an incident on Oct 23, and on Oct 24 the attacker used it to reach Fireworks development machines and run an unauthorized workload. Fireworks says the evidence links a financially motivated group that sells stolen LLM access to illicit routing gateways, and that it has seen no further successful access since Oct 25."
status: "official (Fireworks AI incident update, Nov 9, 2026; original bulletin Oct 9-10); investigation ongoing with Palo Alto Networks Unit 42"
published: 2026-11-09
---

> Fireworks first disclosed on Oct 9 that internal credentials were used to read secrets in evaluation-job environment variables. Its Nov 9 update says the affected internal application also exposed resource metadata and billing information (addresses, invoices, balances, credit usage, but no card numbers), and that for a limited set of accounts secrets, datasets and incomplete model weights were accessed. A separate path followed: a third-party cloud provider hosting Fireworks R&D machines had an incident on Oct 23, and on Oct 24 the attacker used it to reach Fireworks development machines and run an unauthorized workload. Fireworks says the evidence links a financially motivated group that sells stolen LLM access to illicit routing gateways, and that it has seen no further successful access since Oct 25.

[← AI](/ai) · [markdown](/items/fireworks-ai-incident-update-llmjacking-billing-datasets-incomplete-model-weights.md)

official (Fireworks AI incident update, Nov 9, 2026; original bulletin Oct 9-10); investigation ongoing with Palo Alto Networks Unit 42

2026-11-09

### Fireworks AI widens its October breach: billing data, and for some customers secrets, datasets and incomplete model weights; it ties the actor to LLMjacking

- **Not affected, per Fireworks:** inference, Training SDK and Nexus, which keep no data by default.
- **Service changes:** self-serve dataset uploads and downloads for Batch Inference and dataset-based training are disabled during the investigation.
- **If you're a customer:** follow your account-specific notice; rotate third-party credentials with the providers that issued them, since deleting a stored secret doesn't invalidate it.
- Context: this fits the wave of AI developer-platform intrusions for LLM access covered in our Oct 30 Wiz "Black October" and Northflank items.

Sources: [Fireworks AI: Update on October 2026 security incident (Nov 9, 2026)](https://fireworks.ai/security/bulletins/october-2026-security-incident-update) · [Fireworks AI: October 2026 security incident bulletin (Oct 9-10, 2026)](https://fireworks.ai/security/bulletins/october-2026-security-incident)