INSPECTING ARCHIVED INTELLIGENCE (OUTDATED VERSION).
Fireworks AI Exploited Notion
|
2026-11-12 11:08
CRITICAL
LOW
AI-ENABLED ATTACK · AUTONOMOUS
DATA BREACH
Executive Summary
AI-generated
A data breach involving Fireworks AI occurred in October, with the second incident linked to a recent rise in LLMJacking activity. The attackers accessed Notion's customer AI chat transcripts used for the Insights feature on October 24, compromising sensitive information from fewer than 1% of affected workspaces. This was part of a larger breach at Fireworks AI, which experienced two incidents: one through compromised internal credentials and another via a third-party cloud provider hosting its R&D machines. The attackers accessed environment variables containing API keys and tokens for a limited number of accounts in early October. Notion notified customers on November 2 that their data had been accessed by Fireworks AI, with the company recommending they treat the exposed data as compromised and rotate any exposed credentials.
Technical Mitigations AI-generated
• Review the data through Security incident review in Notion’s desktop web interface and share a link with members that lets them download your own transcripts.
• Rotate any passwords, API keys, or other credentials entered into affected chats.
Intelligence Metadata
Actors / Malware / CVEs / Campaigns
Target & Sectors
APAC
APAC
LATAM
LATAM
NORTH_AMERICA
NORTH_AMERICA
retailretail
financefinance
technologytechnology
transportationtransportation
Incident Timeline
October 6
Fireworks initially discovered a data breach on October 6, and subsequently hired Palo Alto Networks Unit 42 to investigate the incident.
Click on any entity below to view its context and source!
tactic
Data Breach
Fireworks published an updated notice on Monday saying it hired Palo Alto Networks Unit 42 to investigate a data breach it initially
discovered on October 6
.
organisation
Palo Alto Networks Unit
Fireworks published an updated notice on Monday saying it hired Palo Alto Networks Unit 42 to investigate a data breach it initially
discovered on October 6
.
October 10, 2026
Fireworks hackers used stolen internal credentials to access environment variables in a limited number of customer accounts, exposing billing information.
Click on any entity below to view its context and source!
target_region
United States
General Document Context
2026/10/11
Fireworks AI disclosed on Monday that its investigation into a security incident uncovered further malicious activity, affecting account and billing information from a "limited set of accounts" during two breaches last month.
October 23
Threat actors accessed Fireworks AI's development environment via a third-party cloud provider on October 23, compromising billing information and other sensitive data.
Click on any entity below to view its context and source!
organisation
Fireworks AI
The disclosure comes shortly after
Fireworks AI confirmed
that an attacker accessed its development environment via a third-party cloud provider on October 23 and compromised resource metadata, billing information, model weights, and datasets for some customers.
organisation
Fireworks
Monday’s alert from Fireworks also notes that the company experienced another incident on October 23 where a third-party cloud provider that hosted their R&D machines was attacked by cybercriminals.
October 23, 2026
General Document Context The company said attackers reached its development environment via an unnamed third-party cloud provider that hosted Fireworks research and development (R&D) machines on October 23, 2026.
Click on any entity below to view its context and source!
target_region
United States
General Document Context
October 24
An unauthorized third party used Notion to reach Fireworks development machines and then executed an unauthorized workload that led to the compromise of customer billing information.
October 25
As of October 25, we have not detected further evidence of successful unauthorized access by the threat actor on the Fireworks platform customers use.
November 2
Fireworks hackers accessed and stole synthetic test data from Notion AI on November 2, prompting an alert that was later confirmed to contain exposed chat transcripts.
Click on any entity below to view its context and source!
organisation
Notion AI
Fireworks contained the incident the same day and notified Notion on November 2 that Notion AI data had been accessed.
November 3
Fireworks hackers stole data billing info from Notion customer, prompting the company to take drastic measures by cutting off traffic to Fireworks and deleting all vendor data.
November 6
Fireworks hackers used unauthorized access to steal sensitive billing information from Notion customers.
November 8
Fireworks hackers accessed and stole data billing information from Notion customers on November 8.
Click on any entity below to view its context and source!
target_region
United States
“Upon analysis of the dataset Fireworks AI gave us on November 8, we discovered that it included Notion AI chat transcripts used for the Insights feature, which for the vast majority of impacted workspaces were unattributable,”
Notion stated in an update
.
2026/11/12
Threat actors used Fireworks AI to target Notion, stealing data billing info from fewer than 1% of workspaces.
Click on any entity below to view its context and source!
organisation
Notion
SecurityWeek
has emailed Notion for additional information on the data breach and will update this article if the company responds.
Notion has also
published a FAQ page
about the incident, highlighting that Fireworks had contractually committed to zero data retention for Notion data.
"
Notion has also emphasized that no other features across its platform were affected by the incident.
organisation
Notion AI
US - Notion / Fireworks AI
Notion disclosed a third-party data breach involving its AI provider, Fireworks AI, which exposed Notion AI chat transcripts associated with the Insights feature, affecting fewer than 1% of workspaces.
Notion, the digital workspace platform, disclosed that a security breach at its AI infrastructure provider, Fireworks AI, exposed Notion AI chat transcripts in a small number of workspaces.
"
While the initial assessment was that the breach allowed the threat actor to access Notion AI synthetic testing data, it has since come to light that it may also have involved potential customer data.
organisation
LLMjacking
Mandiant said it has responded to multiple LLMJacking incidents in 2026 spanning a variety of cloud and LLM providers while impacting “organizations in North America, South America, and Asia Pacific across the financial services, retail, technology, and transportation sectors.”
One week ago, researchers at Google's Threat Intelligence Group (GTIG) and Mandiant reported that large language model hijacking attacks (also known as LLMjacking) are
driving record costs for compromised organizations
, with cloud AI charges reaching
as high as $1 million due to fraudulent calls using stolen API keys.
organisation
Google
LLMJacking
Fireworks said its breaches are part of a larger string of attacks recently spotlighted by incident responders at Google.
FireWorks provides a high-performance inference platform for developers who want to deploy and run open-source artificial intelligence models (such as Mistral, Meta's Llama, or Google's Gemma) without worrying about hardware bottlenecks and complex infrastructure tasks.
financial
$1 million
One week ago, researchers at Google's Threat Intelligence Group (GTIG) and Mandiant reported that large language model hijacking attacks (also known as LLMjacking) are
driving record costs for compromised organizations
, with cloud AI charges reaching
as high as $1 million due to fraudulent calls using stolen API keys.
“Across individual incidents, we observed fraudulent charges associated with resource consumption exceeding $100,000 in a single day, with cumulative charges at some organizations reaching $1 million before access was revoked.”
infrastructure
Cursor
Related:
Hackers Deploy Malware in Fresh ScreenConnect Campaign
Related:
Maersk Confirms Data Theft Amid $5 Million Extortion Attempt
Related:
Hackers Target N-central Zero-Days as N-able Rushes Patches
Related:
Malicious Claude Code Extension Served in Cursor Supply Chain Attack
financial
$5 Million
Related:
Hackers Deploy Malware in Fresh ScreenConnect Campaign
Related:
Maersk Confirms Data Theft Amid $5 Million Extortion Attempt
Related:
Hackers Target N-central Zero-Days as N-able Rushes Patches
Related:
Malicious Claude Code Extension Served in Cursor Supply Chain Attack
organisation
Adaptive Security's
Adaptive Security's AI agents analyze email intent, context, and behavior to uncover phishing, business email compromise, and invoice fraud.
organisation
Microsoft 365
See what's getting through with a 14-day read-only scan of Microsoft 365 or Google Workspace, with no MX changes.
organisation
Google Workspace
See what's getting through with a 14-day read-only scan of Microsoft 365 or Google Workspace, with no MX changes.
organisation
Fireworks
FireWorks provides a high-performance inference platform for developers who want to deploy and run open-source artificial intelligence models (such as Mistral, Meta's Llama, or Google's Gemma) without worrying about hardware bottlenecks and complex infrastructure tasks.
organisation
Mistral
FireWorks provides a high-performance inference platform for developers who want to deploy and run open-source artificial intelligence models (such as Mistral, Meta's Llama, or Google's Gemma) without worrying about hardware bottlenecks and complex infrastructure tasks.
organisation
Meta
FireWorks provides a high-performance inference platform for developers who want to deploy and run open-source artificial intelligence models (such as Mistral, Meta's Llama, or Google's Gemma) without worrying about hardware bottlenecks and complex infrastructure tasks.
organisation
Gemma
FireWorks provides a high-performance inference platform for developers who want to deploy and run open-source artificial intelligence models (such as Mistral, Meta's Llama, or Google's Gemma) without worrying about hardware bottlenecks and complex infrastructure tasks.
organisation
Mandiant
One week ago, researchers at Google's Threat Intelligence Group (GTIG) and Mandiant reported that large language model hijacking attacks (also known as LLMjacking) are
driving record costs for compromised organizations
, with cloud AI charges reaching
as high as $1 million due to fraudulent calls using stolen API keys.
victims
10,000 customers
The San Mateo-based company was founded in 2022 and has more than 10,000 customers, offering an array of AI systems and tools to companies.
organisation
Palo Alto Networks
Fireworks is working with Palo Alto Networks on hardening its security posture and revoking compromised credentials.
organisation
Digital
Digital workspace company Notion
released
its own notice that some of its customers were impacted by the Fireworks breaches.
organisation
Breaking News
“This operational specialization will sustain a growing illicit resale marketplace, allowing threat actors to rapidly siphon enterprise quotas while passing the billing costs to victims before compromises are detected.”
is a Breaking News Reporter at Recorded Future News.
organisation
Recorded Future News
“This operational specialization will sustain a growing illicit resale marketplace, allowing threat actors to rapidly siphon enterprise quotas while passing the billing costs to victims before compromises are detected.”
is a Breaking News Reporter at Recorded Future News.
organisation
TechRepublic
He previously covered cybersecurity at ZDNet and TechRepublic.
financial
$100,000
“Across individual incidents, we observed fraudulent charges associated with resource consumption exceeding $100,000 in a single day, with cumulative charges at some organizations reaching $1 million before access was revoked.”
organisation
Large Language Models
"The evidence indicates links to a cybercriminal threat actor group, which is financially motivated and aimed at selling tokens from unauthorized access to Large Language Models (LLMs) to illicit routing gateways, known as 'LLMJacking',"
Fireworks said
at the time.
organisation
Palo Alto Networks'
The company is now working with Palo Alto Networks' Unit 42 incident response division to secure its systems against future attacks and further investigate the incident's full scope and impact.
Tactical Metrics
Metrics
infrastructure
Cursor
Affected Product
Click for context!
Related:
Hackers Deploy Malware in Fresh ScreenConnect Campaign
Related:
Maersk Confirms Data Theft Amid $5 Million Extortion Attempt
Related:
Hackers Target N-central Zero-Days as N-able Rushes Patches
Related:
Malicious Claude Code Extension Served in Cursor Supply Chain Attack
Metrics
financial
$5M
Ransom
Related:
Hackers Deploy Malware in Fresh ScreenConnect Campaign
Related:
Maersk Confirms Data Theft Amid $5 Million Extortion Attempt
Related:
Hackers Target N-central Zero-Days as N-able Rushes Patches
Related:
Malicious Claude Code Extension Served in Cursor Supply Chain Attack
Metrics
infrastructure
Microsoft 365
Affected Product
See what's getting through with a 14-day read-only scan of Microsoft 365 or Google Workspace, with no MX changes.
Metrics
victims
10,000
Customers
The San Mateo-based company was founded in 2022 and has more than 10,000 customers, offering an array of AI systems and tools to companies.
Metrics
financial
$100K
Fraud Losses
“Across individual incidents, we observed fraudulent charges associated with resource consumption exceeding $100,000 in a single day, with cumulative charges at some organizations reaching $1 million before access was revoked.”
Metrics
financial
$1M
Financial Impact
“Across individual incidents, we observed fraudulent charges associated with resource consumption exceeding $100,000 in a single day, with cumulative charges at some organizations reaching $1 million before access was revoked.”
One week ago, researchers at Google's Threat Intelligence Group (GTIG) and Mandiant reported that large language model hijacking attacks (also known as LLMjacking) are
driving record costs for compromised organizations
, with cloud AI charges reaching
as high as $1 million due to fraudulent calls using stolen API keys.
Intelligence Sources
BleepingComputer
2026-11-10
Fireworks AI discloses new breach after cloud provider hack
BleepingComputer
TheRecord
2026-11-10
The Hacker News
2026-11-10
Bluesky Hackmanac
2026-11-11
Cyber Alert !! US - Notion / Fireworks AI Notion disclosed...
Bluesky Hackmanac
BleepingComputer
2026-11-11
Notion says Fireworks breach exposed customer AI chat transcripts
BleepingComputer
SecurityWeek
2026-11-12
Unpublish from Social Media?
Are you sure you want to delete this podcast video from all synchronized social networks (YouTube, Facebook, Threads)?
Important:
Due to Meta API restrictions, Instagram Reels cannot be deleted automatically via API by third-party apps.
View Profile to Delete Manually
View Profile to Delete Manually
Tactical Intelligence
Report Intelligence Issue
Podcast Options
Generate
Reset / Delete
Incident Version History
CURRENT VERSION
Last Updated: 2026-11-13T07:09
Comprehensive Tactical Telemetry
Highly Correlated Entities
22x
organisation
Identified Entity
Notion
entity
17x
timeline
Temporal Reference
October 24
date
6x
tactic
Cyber Operation Type
Data Breach
tactic
3x
target region
Target Region
LATAM
region
2x
infrastructure
Affected Product
Cursor
software
2x
attribution
Attributing Entity
Google Threat Intelligence Group
authority
Contextual Telemetry
Context Block
9 METRICS
tactic
MITRE ATT&CK Technique
T1588.001 - Malware
technique
financial
Ransom
$5M
general metric
%
1
%
target region
Target Country
United States
country
general metric
A.M. Pdt
9
a.m. pdt
victims
Customers
10,000
customers
financial
Fraud Losses
$100K
financial
Financial Impact
$1M
general metric
Incident
42
incident
Click on any entity below to view its context in the main text!
Selective Unpublish
Select the networks you want to remove this post from. The system will try to delete the real post through the API and clean the database so you can publish it again.
By navigating this website, you accept the use of strictly necessary technical cookies for session security and basic platform functionality. We do not use tracking or advertising cookies. Read our Privacy Policy.
Generating report…
0s