Return to Threats

Fireworks AI Confirms Wider Breach Affecting Customer Datasets, Model Weights and Notion AI Chats

thehackernews.com 2026-11-10 data leakage Critical

Executive Summary

Fireworks AI disclosed a wider breach affecting customer information and AI assets, including datasets, model weights, incomplete models, secrets, and Notion AI chats. Organizations using Fireworks AI or similar AI platforms should review exposed assets and strengthen controls around sensitive AI data and artifacts.

What Happened

Fireworks AI disclosed on Monday that its investigation into a security incident first reported early last month uncovered further malicious activity, affecting account and billing information, alongside secrets, datasets, and incomplete models for a "limited set of accounts." In a notice published on October 10, 2026, the company said it detected unauthorized use of internal Fireworks

Affected Products

  • Fireworks AI
  • Notion AI

Vulnerability / Attack Technique

Unauthorized use of internal Fireworks AI systems

Technical Mechanism

The summary states that Fireworks AI detected unauthorized use of internal systems and later identified further malicious activity. It does not describe the initial access method, persistence mechanism, privilege level, or exact exfiltration path.

AI Security Category

LLM02:2025 Sensitive Information Disclosure

Relevant Frameworks

LLM02:2025 Sensitive Information DisclosureMITRE ATLAS: CollectionMITRE ATLAS: Exfiltration

RealGround Analysis

Exposure of datasets, model weights, secrets, and AI chats can affect confidentiality across the AI application lifecycle and may create downstream risks for teams that rely on shared AI infrastructure. AI teams should include hosted model providers, agent integrations, training artifacts, and user conversations in breach-readiness and vendor-risk assessments.

Recommended Controls

  • Maintain an inventory of AI datasets, model weights, incomplete models, secrets, user chats, and third-party AI integrations.
  • Apply least-privilege access, strong authentication, and separate administrative access for customer data and model artifacts.
  • Store secrets in a dedicated secrets manager, rotate credentials after suspected unauthorized access, and prevent secrets from being embedded in datasets or model artifacts.
  • Enable centralized logging and alerting for unusual access to datasets, model files, customer conversations, and billing information.
  • Require incident-notification, data-protection, access-control, and deletion commitments from AI platform and integration vendors.

Related RealGround Service

AI Security Readiness Assessment · or start with the free AI Agent Security Scanner.

Sources

thehackernews.com: https://thehackernews.com/2026/11/fireworks-ai-breach-widens-to-expose.html

Summary and analysis written by RealGround from the source above; last updated 2026-11-11.

Talk to AI CISO