🔍 Read the full analysis: Fireworks AI Breach: Review Your Data And Model Access on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Fireworks AI’s November 9 update says an October security incident reached account and billing information and, for a limited set of accounts, secrets, datasets and incomplete model weights. Notion says transcripts from its AI Insights feature may have been in an accessed dataset, affecting fewer than 1% of workspaces. The number of affected Fireworks customers and the fate of any accessed data remain unclear.
Fireworks AI said on November 9 that an October security incident reached further into customer information than it had initially reported, exposing account and billing data and, for a limited set of accounts, secrets, datasets and incomplete models. Notion separately said transcripts from its Notion AI Insights feature may have been included in a dataset attackers could access, affecting fewer than 1% of workspaces.
Fireworks’ first bulletin, published October 9, described misuse of internal credentials to read environment variables from evaluation jobs belonging to a limited number of customer accounts. At that point, the company said it had not detected unauthorized access to customer models, datasets, training data, inference traffic or personal information. Its November 9 update revised that account: the investigation had since found impacts to account and billing information and, for some accounts, secrets, datasets and incomplete model weights.
The company lists potentially accessed account information as resource identifiers, display names, descriptions and creation times. Billing information included billing and shipping addresses, invoice and payment history, account balances and credit usage. Fireworks said a subset of incomplete model weights and datasets was accessed for certain customers; it has not specified how many customers or described what “incomplete” means for the models.
Fireworks described two routes into its systems. In October, someone used internal credentials to read environment variables from customer evaluation jobs. Later, after a security incident at a third-party cloud provider hosting Fireworks research and development machines, an unauthorized party reached development machines and ran a workload on October 24, the company said. Fireworks says it detected that activity the same day. Notion’s advisory says Fireworks informed it on November 6 that potential customer data may have been accessed and confirmed the dataset on November 8.
Fireworks said it does not store payment card numbers and that they were not affected. It also reported no detected impact to its Inference, Training SDK or Nexus products, which it says are zero-data-retention by default. The company disabled self-service dataset uploads and downloads used by Batch Inference and dataset-based training while it investigates.
Fireworks AI breach Review data and model access
A November update expanded the reported scope of an October incident. Customers should review account notices and check whether their data or credentials may be affected.
What information was involved
Reported access varied by account; the number of affected customers remains unknown.
Identifiers and billing
Account details and billing records may have been accessed. Fireworks says payment card numbers were not stored or affected.
Secrets and datasets
Fireworks reported access to secrets, datasets and incomplete model weights for a limited set of accounts.
Transcripts may be included
Notion said transcripts may have been in an accessed dataset. This does not establish that every transcript was read.
How the reported scope changed
Fireworks first described credential misuse, then reported additional access after a later cloud-provider incident.
- Internal credentials used to read environment variables from evaluation jobs.
- No detected access then reported to customer datasets, models or personal information.
- Account and billing information, plus some secrets, datasets and incomplete weights, reported accessed.
- A workload ran on a development machine on October 24.
What customers can do now
Use account-specific notices to assess exposure while the investigation continues.
Review messages from Fireworks and Notion for affected resources or accounts.
Assess relevant datasets, secrets and model materials against your account details.
Fireworks says its investigation is ongoing and self-service dataset transfers remain disabled.
Risks for AI Platform Customers
The reported access matters because businesses often send proprietary material to AI platforms to train or evaluate models. Datasets, secrets and model weights can contain sensitive business information or help protect access to other systems. Fireworks has not said what the accessed data contained in individual cases, so the practical exposure for each customer cannot yet be determined.
Notion’s disclosure gives one concrete example of possible downstream impact: it says chat transcripts from Notion AI Insights were in a dataset attackers may have accessed. Notion put the share of affected workspaces below 1%, but did not provide a total number in the source material. The disclosure underscores that a breach at an infrastructure provider can affect information handled for customers using separate products.
The incident also changes the scope customers must assess. Fireworks’ initial bulletin said it had not detected access to several categories, including datasets and personal information; the later update reports access to some datasets and account information. That does not establish that every customer or product was affected. It does mean customers should use Fireworks’ account-specific notices, where available, to determine whether their resources or credentials are implicated.
Products Worth Considering
Dell UltraSharp U2424H 24-Inch FHD Monitor
Dell UltraSharp U2724DE 27-Inch WQHD IPS Black Monitor
Logitech MX Ergo S Advanced Wireless Trackball Mouse USB-C – Graphite
As an affiliate, we earn on qualifying purchases.
From October Alert to November Update
Fireworks said it identified activity and began containment at 9 a.m. Pacific time on October 6, then began notifying affected customers the following morning. On October 9, it published its first bulletin and said it had rotated employee API keys. It also revoked additional user API keys and rotated service-account tokens on October 10, according to its account of events.
The later cloud-provider incident occurred on October 23, Fireworks said. The unauthorized workload ran on October 24 and led to additional customer data impact. The company said it has detected no further evidence of successful unauthorized access by the threat actor on the customer-facing Fireworks platform since October 25. It published the expanded account on November 9, 34 days after its reported initial detection.
Fireworks said it notified law enforcement and retained Palo Alto Networks Unit 42 to conduct forensic work. It has not named the third-party cloud provider or a suspected threat group in the supplied account. The company’s description of the evidence as indicating links to a cybercriminal group is a company assessment, not an identified attribution.
“For a limited set of accounts, we also discovered impact to secrets, datasets, and incomplete models.”
— Fireworks AI, November 9 security update
Affected Accounts and Data Still Unknown
Fireworks has not disclosed the number of affected customers or accounts, nor identified the third-party cloud provider involved. The company has also not explained how the internal credentials were first obtained, what the attackers did with any accessed information, or whether the data was copied, retained, published or used. Those details remain unconfirmed in the material available.
The meaning and extent of the reported access to incomplete model weights are also unclear. Fireworks has not said whether these were partial copies or unfinished training outputs, or described what information any particular dataset contained. Notion’s statement concerns a dataset that may have been accessed; it does not establish that every transcript in the dataset was read or that all Notion AI Insights users were affected.
Fireworks says it has not detected further successful unauthorized access to the customer platform since October 25, but its investigation remains open. That statement describes what the company has detected so far; it does not settle what happened to data already accessed or rule out later discoveries.
Forensics and Customer Notifications Continue
Fireworks says it is continuing its investigation with help from Palo Alto Networks Unit 42 and has notified law enforcement. It has not provided a date for a final report or said when it will publish more detail. Its update says the investigation may identify additional information, so customers should watch for further notices from the company.
While the review continues, Fireworks says self-service dataset uploads and downloads for Batch Inference and dataset-based training are disabled. The company has not stated when those functions will return. It also has not described a schedule for restoring them or provided a public estimate of how many customers may need to take action.
Customers can check communications from Fireworks and Notion for account-specific information and review whether credentials, datasets or model resources are covered by those notices. The source material does not establish that every customer needs to rotate credentials or that any particular data was misused. Further company updates and findings from the forensic investigation are the next milestones that may clarify the scope and consequences.
Key Questions
What information does Fireworks say was affected?
Fireworks reports access to account and billing information and, for a limited set of accounts, secrets, datasets and incomplete model weights. It has not disclosed the number of affected accounts.
Were payment card numbers exposed?
Fireworks says it does not store payment card numbers and that they were not affected. This is the company’s statement in its November 9 update.
What did Notion say about its users?
Notion said transcripts from its Notion AI Insights feature were in a dataset attackers may have accessed. It reported that fewer than 1% of workspaces were affected, without specifying a total number in the source material.
Has Fireworks said who was responsible?
No group or cloud provider was named in the supplied update. Fireworks said evidence indicated links to a cybercriminal group, but the group’s identity and how the internal credentials were obtained remain unclear.
What happens to dataset uploads and downloads now?
Fireworks says self-service uploads and downloads used by Batch Inference and dataset-based training are disabled during the investigation. It has not announced when those functions will be restored.
Source: ThorstenMeyerAI.com
