Fireworks AI says breach exposed billing data and incomplete model assets

Fireworks AI has expanded its account of a security incident first disclosed in October, saying the intrusion affected account and billing information as well as secrets, datasets and incomplete model weights for a limited set of customer accounts. The AI infrastructure provider said payment card numbers are not stored by Fireworks and were not affected.
The company first detected suspicious activity on October 6, 2026, at 9 a.m. PDT. Its October 10 notice said an unauthorized third party had used internal Fireworks credentials to access environment variables used in evaluation jobs in a limited number of customer accounts. Those variables could include API keys and tokens held as secrets.
At that stage, Fireworks said it had found no evidence of access to customer models, datasets, training data, inference traffic or personal information. Its latest bulletin revises the known scope after additional activity was identified in the development environment.
Development systems were used for additional access
Fireworks said an attacker reached its development environment through an unnamed third-party cloud provider hosting the company’s research and development machines on October 23. On October 24, the unauthorized party used that access to reach Fireworks development machines and run an unauthorized workload, leading to further customer data impact.
As of October 25, Fireworks said it had not detected further evidence of successful unauthorized access on the platform customers use. The company is notifying affected customers.
Data categories and containment measures
The newly identified exposure can include resource metadata, such as resource identifiers, display names, descriptions and creation times. It can also include billing and shipping addresses, invoice and payment history, account balances and credit usage.
For certain customers, the incident also involved a subset of incomplete model weights and datasets. The company did not state that complete customer models or payment card numbers were exposed.
- Fireworks revoked compromised internal credentials.
- It rotated service-account tokens, affected Fireworks API keys and keys created by the threat actor.
- It isolated affected systems to limit further unauthorized activity and lateral movement.
- It tightened access controls and improved logging and monitoring.
LLMjacking is the stated motive
Fireworks said the available evidence points to a financially motivated cybercriminal group seeking to sell tokens obtained through unauthorized access to large language models. The activity is assessed as LLMjacking: the theft or misuse of credentials to hijack LLM capacity and resell access through illicit routing gateways, potentially leaving the legitimate owner with the resulting charges.
Google warned last week that attackers are increasingly targeting AI infrastructure for LLMjacking across financial services, retail, technology and transportation in North America, South America and Asia Pacific. For businesses using AI infrastructure, the incident makes evaluation-job secrets, third-party cloud access and billing records concrete areas to include in credential rotation, access-control review and monitoring processes.

