🔍 Read the full analysis: Fireworks AI Breach: Seven Areas AI Customers Can Check on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Fireworks AI’s November 9 update says an incident first disclosed in October affected more data than initially reported, including account and billing information and, for a limited set of accounts, secrets, datasets and incomplete models. Notion says chat transcripts from its AI Insights feature may have been in an accessed dataset, affecting fewer than 1% of its workspaces. The number of affected customers and the fate of any accessed data remain unclear.
Fireworks AI said on November 9 that an October security incident exposed more customer information than it had initially reported, including account and billing details and, for a limited set of accounts, secrets, datasets and incomplete models. The update matters to companies using Fireworks to host or fine-tune AI models because the platform handles customer data and development resources; the company has not disclosed how many customers were affected.
Fireworks’ October 9 bulletin said an attacker had used internal company credentials to read environment variables from a limited number of customer evaluation jobs. Those variables can contain settings such as API keys or tokens. At the time, Fireworks said it had not detected access to customer models, datasets, training data, inference traffic or personal information. Its November 9 update added that the incident affected account and billing information, and that some accounts’ secrets, datasets and incomplete models were accessed.
Fireworks described two stages of activity. It said it identified the initial activity on October 6 and began containment. On October 23, a third-party cloud provider hosting Fireworks research and development machines experienced a security incident, according to Fireworks. The company says an unauthorized party used that access on October 24 to reach development machines and run an unauthorized workload that caused additional customer data impact. Fireworks said it detected that activity the same day and has not detected further successful access by the actor on the customer platform since October 25.
The company listed exposed information as resource metadata, billing details and, for certain customers, a subset of incomplete model weights and datasets. It says payment card numbers were not affected because Fireworks does not store them. It also said it has not detected an impact to its Inference, Training SDK or Nexus products, which it describes as zero-data-retention by default. Fireworks has disabled self-service dataset uploads and downloads used by Batch Inference and dataset-based training while its investigation continues.
Fireworks AI breach Seven areas customers can check
Fireworks AI’s November 9 update expanded its account of an October incident. The affected customer count and the fate of accessed data remain unclear.
What the update says was accessed
Reported impact varied by account; Fireworks has not published the total number affected.
Account and billing details
These details may reveal information about a customer’s relationship with the service.
Secrets and credentials
Fireworks has not said which secrets were accessed or whether any were used elsewhere.
Datasets and incomplete models
The company has not explained what “incomplete” means for affected model weights.
Notion AI Insights transcripts
Notion said transcripts may have been in a dataset attackers could access; it did not confirm each transcript was viewed.
How the disclosure unfolded
Fireworks described initial activity and later access through a third-party cloud provider’s incident.
Seven checks for customers
Use the disclosures to guide account-specific follow-up; they do not establish that every customer was affected.
Confirm whether Fireworks contacted your account.
Look for information that may have been exposed.
Check credentials available to affected jobs or datasets.
Ask Fireworks whether your account, datasets or models were involved.
Ask whether exposed credentials require action and whether they remained active.
Notion AI Insights users can consult Notion about possible transcript exposure.
Fireworks disabled self-service dataset uploads and downloads during its investigation.
Risks for Fireworks AI Customers
The reported access could matter beyond ordinary account administration. Billing records and account details may reveal information about a customer’s relationship with the service, while secrets such as API credentials can be sensitive if they remain active or are reused. Fireworks has not said which specific secrets were accessed or whether any were used outside the incident.
For customers fine-tuning models, datasets and model weights can contain commercially sensitive material. Fireworks calls the affected weights incomplete, but has not explained what portion of a model or training process they represent. That leaves customers without enough public detail to judge the potential value or usability of the accessed material. The reported Notion exposure also shows that data sent to an AI platform can involve information belonging to users of a separate service.
Notion said chat transcripts from its Notion AI Insights feature were in a dataset that attackers may have accessed. Notion put the share of affected workspaces at fewer than 1%. That is a company-reported proportion, not a count of affected users or transcripts, and the public information does not establish that every transcript in the dataset was viewed or taken.
Products Worth Considering
LG UltraGear 25G590B-B 25-Inch Full HD 1000Hz Gaming Monitor
Samsung Galaxy Tab S7 11-inch Android Tablet 128GB Wi‑Fi with S Pen Mystic Black
Logitech MX Master 4 for Mac Wireless Ergonomic Mouse, White Silver
As an affiliate, we earn on qualifying purchases.
From October Alert to Wider Disclosure
The disclosure developed over several weeks. Fireworks said it identified the initial activity on October 6, began notifying affected customers on October 7 and published its first security bulletin on October 9. It also said it rotated employee API keys on October 9 and revoked additional user API keys and rotated service-account tokens on October 10.
The second stage followed the reported cloud-provider incident on October 23 and unauthorized activity on Fireworks development machines on October 24. Fireworks said it detected that activity that day. Notion’s advisory says Fireworks first contacted it on November 2 about a small dataset believed to contain synthetic test data; on November 6, Fireworks said customer data might have been accessed, and confirmed the dataset on November 8. Fireworks published its expanded account on November 9.
Fireworks said it notified law enforcement and engaged Palo Alto Networks Unit 42 for forensic work. The company’s update does not name the cloud provider or identify a responsible group. It said evidence indicated links to a cybercriminal, but the available material does not establish the actor’s identity or motives.
“For a limited set of accounts, we also discovered impact to secrets, datasets, and incomplete models.”
— Fireworks AI, November 9 security update
Customer Scope Still Undisclosed
Fireworks has not published a total number of affected customers or accounts, nor identified the third-party cloud provider involved. Its description does not explain how the internal credentials used in the initial activity were first obtained, which exact secrets or datasets were accessed, or what “incomplete” means for the model weights.
It is also not clear whether accessed information was copied, used or shared beyond the unauthorized activity Fireworks described. The company has not publicly detailed whether credentials or other exposed secrets were found to remain active after its response, or whether all affected customers have been contacted. Its statement that it has detected no further successful access since October 25 describes the company’s current findings, not a complete public accounting of what happened.
Notion’s workspace percentage does not specify how many transcripts or individuals were involved. The available statements say the transcripts were in a dataset attackers may have accessed; they do not confirm that attackers read each transcript. Fireworks says its investigation is ongoing, so the confirmed scope may change.
Investigation and Customer Actions
Fireworks says it is continuing its investigation with Unit 42 and has disabled certain self-service dataset functions during that work. Further company updates or customer notices may clarify the affected accounts, data and systems. The company has not announced a date for a final report or for restoring the disabled functions.
Customers can use the disclosures as a prompt to check whether their accounts received a direct notice from Fireworks, review account and billing records, and identify any secrets or credentials that may have been available to affected jobs or datasets. They can also ask Fireworks whether their specific resources were involved and whether any exposed credentials require action. These are precautionary checks; the public update does not establish that every Fireworks customer was affected or that every exposed credential was misused.
Organizations using Notion AI Insights can consult Notion’s advisory and contact the company to determine whether their workspace is among those it identified. Until Fireworks publishes more detail, customers will not be able to assess the incident’s full scope from the public statements alone.
Key Questions
What did Fireworks AI say was accessed?
Fireworks reported access to account and billing information and, for a limited set of accounts, secrets, datasets and incomplete models. It also listed resource metadata among the affected information.
Were payment card numbers exposed?
Fireworks says payment card numbers were not affected because the company does not store them. This is the company’s statement in its November 9 update.
Was Notion customer data involved?
Notion said chat transcripts from Notion AI Insights were in a dataset attackers may have accessed. It reported that fewer than 1% of workspaces were affected, without giving a count of transcripts or users.
How many Fireworks customers were affected?
Fireworks has not disclosed a total. Its update says only that a limited set of accounts had impact to secrets, datasets and incomplete models.
What should customers do now?
Customers can check for direct notices from Fireworks, review relevant account and billing records, and ask the company whether their resources or credentials were involved. These checks do not mean that every customer was affected.
Source: ThorstenMeyerAI.com
