Fireworks AI reports intrusion that exposed customer secrets, billing data and some datasets
What happened
Fireworks AI, an LLM inference and fine-tuning platform, disclosed an intrusion in early October 2026 in which an unauthorized party read stored customer secrets for a limited number of accounts. A later update described a second incident tied to a third-party cloud provider that exposed account and billing data and, for some customers, datasets and incomplete model weights. Fireworks links the activity to a financially motivated group and has not given a customer count.
The first bulletin described access by an unauthorized party to an internal application, with stored secrets used in customer evaluation jobs among the data read. The November 9 update says the company later found more. A third-party cloud provider that hosts Fireworks research and development machines had its own security incident on October 23. On October 24 the unauthorized party used that access to reach Fireworks development machines and run a workload, which led to additional customer data impact. Fireworks says it detected this the same day and reports no further unauthorized access to the customer platform as of October 25. The data involved includes resource metadata such as IDs, names and descriptions, and billing information such as billing and shipping addresses, invoice and payment history, account balance and credit usage. For certain customers, a subset of datasets and incomplete model weights were accessed. Fireworks says it does not store payment card numbers and that they were not affected, and that its Inference, Training SDK and Nexus services were not impacted. No count of affected customers is given.
How it happened
Unknown: Unknown. No source has explained how the incident happened.
The software involved works with many AI models, so no single model maker was involved. The product was Fireworks AI platform, a type of AI-powered app.
Impact
Data was exposed. Data involved: credentials, financial, personal data, model weights and internal documents. The impact was confined to one organization and its users or customers.
Severity is rated High, calculated from how far the incident reached and how sensitive the data was.
Was the AI at fault?
No. The failure was in the systems, settings or people around the AI, not in the model's behaviour.
Status and timeline
Confirmed by the affected organization, the AI vendor, a regulator or a named security research firm.
It became public on 9 Oct 2026.
Fireworks revoked and replaced credentials and service accounts, removed unnecessary elevated privileges, isolated parts of its network, tightened access to storage and analytics systems and improved logging. It disabled self-serve dataset uploads and downloads for Batch Inference and dataset-based training during the investigation. Palo Alto Networks Unit 42 is carrying out forensics, law enforcement has been notified, and affected customers are being notified.