Fireworks, an AI inference startup, said in a Monday advisory that attackers reached its development machines after a third-party cloud provider hosting those systems was breached on Oct. 23¹.

The company told customers that on Oct. 24 an unauthorized party used that access to run a workload on Fireworks development machines, which it said led to additional customer data impact¹. It said it detected the activity the same day and, as of Oct. 25, had not seen further successful unauthorized access on the platform customers use¹.

Data the company said was accessed included resource metadata such as identifiers, display names, descriptions and creation times; billing information including billing and shipping addresses, invoice and payment history, account balance and credit usage; and incomplete model weights and datasets for some customers¹. Fireworks said attackers could not obtain payment card numbers because it does not store them¹.

In an early-October notice, Fireworks had already told law enforcement and customers that attackers accessed environment variables with API keys and tokens for a limited number of accounts¹. The company then said evidence pointed to a financially motivated group seeking tokens for unauthorized large-language-model access sold through illicit routing gateways, a pattern known as LLMJacking¹. After revoking compromised credentials, it found the Oct. 23 incident¹.

"A third-party cloud provider that hosted Fireworks R&D machines experienced a security incident on October 23," the company said in the Monday advisory, according to BleepingComputer¹.

Fireworks said it revoked and replaced credentials and service accounts, restricted unauthorized network paths, isolated affected systems and tightened access to data storage and analytics¹. It is working with Palo Alto Networks' Unit 42 on the investigation and further hardening¹. The cloud provider was not named in the BleepingComputer report, and independent confirmation of the full scope was not available¹.