TL;DR
Get privacy and security gear delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
Fireworks AI said on November 9 that attackers accessed account and billing information and, for a limited set of accounts, secrets, datasets and incomplete model weights. The company described an October credential incident and a later intrusion through a third-party cloud provider; the number of affected customers and the fate of accessed data remain unknown.
Fireworks’ October 9 bulletin said someone had used internal credentials to read environment variables from a limited number of customer accounts. The company then said it had not detected unauthorized access to customer models, datasets, training data, inference traffic or personal information. Its November update says the investigation later found impact to account and billing information and access to secrets, datasets and incomplete models for some accounts.
Fireworks describes a second route involving a third-party cloud provider that hosted its research and development machines. According to the company, the provider experienced a security incident on October 23. On October 24, an unauthorized party used that access to reach Fireworks development machines and run an unauthorized workload that led to additional customer data impact. Fireworks says it detected that activity the same day and has not detected further successful access by the threat actor to the platform customers use since October 25.
The company lists resource metadata, billing details and certain datasets and incomplete model weights among the potentially accessed information. Billing information includes addresses, invoice and payment history, account balances and credit usage. Fireworks says it does not store payment card numbers and that it has detected no impact to its Inference, Training SDK or Nexus products, which it says are zero data retention by default. Those statements describe the company’s findings so far; its investigation remains ongoing.
Risks for AI Customer Data
The incident matters because companies may send proprietary datasets, credentials and model artifacts to services that host or fine-tune AI models. Access to those materials could expose business information or create risks if secrets such as API credentials were usable elsewhere. The disclosed access to incomplete model weights may also concern customers whose models reflect their own training data or work, though Fireworks has not explained what “incomplete” means or whether the accessed files could be used.
Fireworks’ account also illustrates how an incident at a cloud or development environment can affect customers of a separate AI platform. The company says the second intrusion reached development machines and caused additional customer data impact. Customers need to establish whether their own accounts, data or credentials were involved before deciding what remediation is needed.
Notion said chat transcripts from its Notion AI Insights feature were in a dataset attackers may have accessed. It said fewer than 1% of workspaces were affected. That is Notion’s account of its own exposure; the available information does not establish how many individual users or conversations were involved.
Products Worth Considering
LG UltraGear 32GX850A-B 32-inch 4K UHD OLED Gaming Monitor
Logitech MX Ergo S Advanced Wireless Trackball Mouse USB-C – Graphite
LG UltraGear 27-inch QHD OLED Gaming Monitor, 240Hz
As an affiliate, we earn on qualifying purchases.
From October Disclosure to Update
Fireworks says it identified the first activity on October 6 and began containment. It says customer notifications began October 7; on October 9, it published its first bulletin and rotated employee API keys. On October 10, it revoked more user API keys and rotated service-account tokens. The company later identified the separate cloud-provider incident, which it dates to October 23, and the resulting activity on Fireworks development machines on October 24.
Notion’s advisory describes a separate notification timeline: Fireworks contacted it on November 2 about a dataset initially understood to contain synthetic test data, then said on November 6 that malicious actors may have accessed potential customer data. Notion says the dataset was confirmed on November 8. Fireworks published its expanded account the next day. It says it notified law enforcement and engaged Palo Alto Networks Unit 42 for forensic work.
The initial bulletin and later update differ in scope. Fireworks first reported no detected access to customer models, datasets, training data, inference traffic or personal information; it later disclosed access to certain datasets, incomplete model weights, secrets and account and billing information. The new findings came as the company’s investigation continued.
“For a limited set of accounts, we also discovered impact to secrets, datasets, and incomplete models.”
— Fireworks AI, November 9 security update
Scope and Data Fate Remain Unknown
Fireworks has not disclosed the number of affected customers or named the third-party cloud provider. Its public account does not explain how the internal credentials used in the first activity were obtained, which specific customers had datasets or incomplete model weights accessed, or what “incomplete” means in practice.
It is also not clear whether attackers copied, retained, published or otherwise used the accessed data. Notion has identified a dataset containing its AI Insights chat transcripts as potentially accessed, but the information provided does not specify how many workspaces or conversations were involved beyond its statement that fewer than 1% of workspaces were affected. Fireworks says its investigation may reveal additional information.
Customer Notifications and Investigation
Fireworks says it began notifying affected customers on October 7 and later revoked additional user API keys and rotated service-account tokens. Customers using the platform can check notices from Fireworks for account-specific details and determine whether any exposed secrets require rotation in connected systems. The company says forensic work is being conducted with Palo Alto Networks Unit 42 and that law enforcement has been notified.
The next public developments depend on the ongoing investigation. Fireworks may provide further findings about affected accounts, the third-party provider and the accessed data. Notion customers who receive a notice can consult Notion’s advisory for information about the workspace data at issue. No further publication date or final incident report has been announced in the source material.
Key Questions
What information does Fireworks AI say attackers accessed?
Fireworks says attackers accessed account and billing information and, for a limited set of accounts, secrets, datasets and incomplete model weights. It also lists resource metadata among the information available through the affected internal application.
How did the intrusions happen?
Fireworks describes an initial October incident involving internal credentials used to read customer job environment variables. It says a second intrusion followed a security incident at a third-party cloud provider hosting Fireworks research and development machines.
Were payment card numbers or Fireworks’ main services affected?
Fireworks says it does not store payment card numbers and that they were not impacted. It also says it has detected no impact to Inference, Training SDK or Nexus, which it describes as zero data retention by default.
Were Notion AI users affected?
Notion said chat transcripts from its AI Insights feature were in a dataset attackers may have accessed. It said fewer than 1% of workspaces were affected; the disclosed material does not give a count of users or conversations.
What remains unknown about the breach?
The public information does not identify the cloud provider, state how many Fireworks customers were affected, explain how the internal credentials were obtained or establish what happened to accessed data. Fireworks says its investigation is ongoing.
Source: Thorsten Meyer AI
