The latest developments in the Fireworks AI cybersecurity incident raise an important question for enterprises increasingly adopting artificial intelligence: How secure is the confidential information employees share with AI-powered applications?

What initially appeared to be a breach involving API credentials and authentication tokens has now expanded to include customer billing information, datasets, incomplete AI model weights and AI chat transcripts processed for Notion.

Notion has confirmed that fewer than 1% of its customer workspaces were affected through its integration with Fireworks AI. Although Notion's own infrastructure was not compromised, some conversations processed through its AI Insights feature may have been exposed.

What makes this development particularly concerning is that a cybersecurity incident at one AI infrastructure provider can potentially expose information belonging to customers of entirely different applications.

Employees increasingly use AI assistants to analyse documents, summarise meetings, discuss business strategies, review source code and process sensitive information. These interactions may contain confidential business information that organisations would never intentionally make publicly accessible.

Yet many users may not realise that their AI conversations can pass through multiple external services before a response is generated.

For enterprises, this raises a critical question: Do organisations know where their AI conversations are processed, how long they are retained and which third-party providers can access them?

The incident reinforces the importance of Data Loss Prevention, Zero Trust access, API security, data minimisation and continuous monitoring of third-party AI infrastructure.

Organisations must establish clear policies governing what information employees can share with AI applications. Sensitive credentials, customer records, proprietary source code and confidential business information should be subject to appropriate restrictions and safeguards.

Equally important is assessing the security practices of AI service providers, including data retention, access controls, encryption, incident response and the use of downstream infrastructure partners.

The larger cybersecurity lesson is that protecting an AI application is not enough. Enterprises must protect the entire chain of services that process their information.

Because when employees share confidential information with an AI assistant, they may also be placing that information in the hands of infrastructure providers they have never heard of.

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

Source: Fireworks AI Breach Widens to Expose Customer Billing Data and Incomplete Model Weights via The Hacker News — published 10 Nov 2026.