Fireworks AI said Monday it found additional unauthorized access to customer data, in an update to a security bulletin first published on Oct. 9. Bloomberg Law described the company as Nvidia-backed and said Fireworks notified customers of a security breach, “its second known hack in a matter of weeks.”
A third-party cloud provider that hosted Fireworks research-and-development machines had a security incident on Oct. 23, the company said. On Oct. 24, an unauthorized party used that access to reach Fireworks development machines and run a workload that led to additional customer data impact, Fireworks said. It said it detected the activity the same day and that, as of Oct. 25, it had not detected further successful unauthorized access on the platform customers use.
Since the first bulletin, Fireworks said, it discovered impacts to account and billing information, including addresses, invoice and payment history, account balances and credit usage. Payment card numbers are not stored by Fireworks and were not impacted, the company said. For a limited set of accounts, it also found that secrets, datasets and a subset of incomplete model weights were accessed.
The earlier bulletin covered an incident Fireworks identified on Oct. 6, in which an unauthorized party used internal credentials to access environment variables, including stored API keys and tokens, in a limited number of customer accounts. At that time, the company said it had not detected access to customer models, datasets, training data, personal information or inference traffic.
Fireworks said the evidence indicates links to a financially motivated group aimed at selling tokens from unauthorized access to large language models through illicit routing gateways, a practice known as LLMJacking. In a Nov. 2 report that Fireworks cited, Google’s Threat Intelligence Group and Mandiant described LLMJacking as hijacking enterprise access to those models and reselling it through illicit gateways, leaving the victim organization to pay the usage charges.
The company said it has not detected any impact to its Inference, Training SDK and Nexus services, which it said are zero-data-retention by default, and that its investigation is ongoing. A Fireworks spokesperson told Bloomberg Law the company is “actively working with our customers to manage the incident, including Notion, Cursor and others.” Bloomberg, citing the company’s website, said Fireworks has more than 10,000 customers, including Notion, Cursor and Harvey. The bulletin does not say which customers were affected.
Sources
- Bloomberg Law: Nvidia-Backed Fireworks AI Discloses Its Second Hack in Weeks (November 9, 2026)
- Fireworks AI: Update on October 2026 security incident (updated November 9, 2026)
- Fireworks AI: Fireworks October 2026 security incident (last updated October 10, 2026)
- Google Threat Intelligence Group and Mandiant: The Rise of LLMJacking (November 2, 2026)
- AI Tech Daily: Binance says attackers used a fake job interview and an internal AI tool in Oct. 29 attack (November 4, 2026)
- AI Tech Daily: Meta patches Muse macOS zero-day that hijacked dictation endpoint (September 22, 2026)