What happened. Fireworks first disclosed on Oct 9 that internal credentials were misused starting Oct 6, exposing environment variables such as API keys and tokens in some customers' evaluation jobs. Its Nov 9 update says a third-party cloud provider that hosted Fireworks R&D machines was breached on Oct 23, and on Oct 24 the attacker used that access to reach Fireworks development machines and run an unauthorized workload. Fireworks says it has seen no further successful access to its customer platform since Oct 25.
What was exposed. An internal application gave access to resource metadata, billing details (billing and shipping addresses, invoice and payment history, account balance and credit usage) and, for certain customers, a subset of incomplete model weights and datasets. Fireworks says it does not store card numbers. It has not said how many customers were affected. The Record reports Fireworks has more than 10,000 customers and that Notion published its own notice saying some of its customers were affected and that it disconnected from Fireworks.
Who did it. Fireworks ties the activity to a financially motivated group that resells stolen access to large language models through illicit routing gateways, a practice called LLMjacking. Palo Alto Networks Unit 42 is handling the forensics.
Why this matters to an enterprise. Fine-tuned model weights and training datasets are now breach data, and an AI vendor's own cloud suppliers sit in your supply chain. If your teams use Fireworks, rotate any keys and secrets stored there, check for usage you don't recognize, and ask the vendor in writing which of your models or datasets were touched.