TL;DR
Get everyday essentials 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’s November 9 update says attackers accessed account and billing information and, for a limited set of accounts, secrets, datasets and incomplete models. Customers should review Fireworks’ notices, check what information and credentials were connected to affected services, and contact the company if they need account-specific details.
Fireworks AI said on November 9 that attackers accessed account and billing information and, for a limited set of accounts, secrets, datasets and incomplete models during security incidents in October. The update broadens the company’s October 9 account of the breach and gives customers reason to check notices from Fireworks and assess which data or credentials they may have stored with the platform.
Fireworks said the incidents involved two routes of access. In the first, someone used what the company called “internal Fireworks credentials” to read environment variables from customers’ evaluation jobs. Those variables can contain settings and credentials used by software. The company’s October bulletin said it had not detected access to customer models, datasets, training data, inference traffic or personal information.
Fireworks later identified a separate incident 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 the access to reach Fireworks development machines and run a workload that led to additional customer data impact. Fireworks said it detected this activity that day.
In its November update, Fireworks listed resource metadata and billing information among the data accessible through an affected internal application. Billing details could include addresses, invoice and payment history, account balances and credit usage. It said a subset of incomplete model weights and datasets was accessed for certain customers. The company did not say how many customers were affected.
What Customers Should Check
The reported exposure matters because Fireworks provides infrastructure for companies that run or fine-tune AI models. Customer evaluation jobs, datasets and model weights can contain material tied to a company’s products or operations. The update does not establish that every listed type of data was accessed for every customer; Fireworks describes the impact to secrets, datasets and incomplete models as applying to a limited set of accounts.
Customers can start by checking their Fireworks account and any security or incident notices sent by the company. They should identify whether evaluation jobs or other affected resources contained API keys, tokens or other secrets, and follow their organization’s process for replacing credentials that may have been exposed. They can also review what datasets and model artifacts were associated with those jobs and whether billing or contact information needs attention.
Notion said chat transcripts from its Notion AI Insights feature were in a dataset attackers may have accessed, and that fewer than 1% of its workspaces were affected. That is a Notion-specific figure; it does not establish the overall number of Fireworks customers or accounts involved. Customers using other services should seek information from those providers or directly from Fireworks.
Products Worth Considering
Apple AirPods Pro (2nd Generation) Wireless Earbuds with USB‑C Charging, Up to 2X More Active Noise Cancelling, Transparency Mode, Adaptive
Logitech MX Ergo S Advanced Wireless Trackball Mouse USB-C – Graphite
Samsung Galaxy Tab A11+ 8GB RAM, 256GB Storage, Gray
As an affiliate, we earn on qualifying purchases.
From October Bulletin to Update
Fireworks said it identified activity on October 6 and began containment. It said it started notifying affected customers on October 7, then published its initial bulletin on October 9 after rotating employee API keys. The company said it revoked more user API keys and rotated service-account tokens on October 10.
The later cloud-provider incident changed the scope described publicly. Fireworks said it detected the unauthorized workload on October 24 and had not detected further successful access by the actor on the customer-facing platform after October 25. The company published the expanded account on November 9.
Notion described receiving information from Fireworks between November 2 and November 8. It said an initial dataset was identified as synthetic test data, then it was told that malicious actors may have accessed potential customer data. Notion said Fireworks confirmed the dataset on November 8. Fireworks said it notified law enforcement and engaged Palo Alto Networks Unit 42 for forensic work.
““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 Use Remain Unknown
Several details remain unavailable in the source material. Fireworks has not named the third-party cloud provider, said how many customers or accounts were affected, or explained how the internal credentials used in the first incident were obtained. It also has not said what happened to data the attackers accessed.
The company describes some model weights as “incomplete” but does not explain whether that means unfinished training runs or partial copies. Its update says its investigation is ongoing and may reveal more information. Fireworks has also said it did not detect impact to Inference, Training SDK and Nexus, which it describes as zero data retention by default; that statement reflects the company’s findings so far.
Notion’s statement concerns its own workspaces and the dataset containing its transcripts. It should not be read as a count of all affected Fireworks customers. The available information does not establish whether other customer datasets contained prompts, outputs or personal information.
Watch for Customer Notices
Fireworks says its investigation is continuing and that it may disclose additional information. Customers can watch for direct notices and follow-up guidance from Fireworks, and ask the company whether their accounts, resources or data were involved. Organizations that find potentially exposed credentials should follow their own incident-response procedures and coordinate with the teams responsible for those credentials.
Further updates may clarify the affected customer count, the cloud provider involved and the nature of the incomplete model artifacts. Until those details are provided, customers should rely on account-specific information from Fireworks and avoid treating the limited figures disclosed by Notion as the overall scope of the incident.
Key Questions
What did Fireworks AI say attackers accessed?
Fireworks said attackers accessed account and billing information and, for a limited set of accounts, secrets, datasets and incomplete models. It also listed resource metadata among the information accessible through an affected internal application.
Were payment card numbers exposed?
Fireworks said it does not store payment card numbers and that they were not impacted. The company did report access to other billing information, including addresses, invoice and payment history, account balances and credit usage.
Did the breach affect Notion AI Insights?
Notion said chat transcripts from its AI Insights feature were in a dataset attackers may have accessed. It said fewer than 1% of its workspaces were affected; that figure applies to Notion, not all Fireworks customers.
What should Fireworks customers do now?
Check for notices from Fireworks, determine whether affected jobs or resources contained credentials, datasets or model artifacts, and ask Fireworks for account-specific information. If a credential may have been exposed, follow the organization’s process for replacing it.
How many Fireworks customers were affected?
Fireworks has not disclosed a total customer or account count in the material available. Its description of a “limited set” for some data categories does not provide a precise overall scope.
Source: Thorsten Meyer AI
