Cyberattack on Fireworks AI: Notion Customers’ Private Chat Transcripts Exposed

20:30

A cybersecurity incident involving artificial intelligence infrastructure provider Fireworks AI has exposed sensitive customer information, including conversations generated through Notion’s AI services. The incident highlights the growing security challenges facing technology companies that depend on external providers to operate their digital products.

The breach occurred after attackers exploited a security incident involving a third-party cloud provider used by Fireworks AI. This access reportedly enabled them to penetrate a development environment belonging to the AI infrastructure company and obtain information associated with several customers.

Among the compromised material were billing records, customer datasets and other technical information. The incident also extended beyond Fireworks AI’s immediate clients, demonstrating how vulnerabilities within interconnected technology services can expose information belonging to users of entirely different platforms.

Notion Confirms Exposure of AI Conversations

Notion confirmed that some of its customers were affected after discovering that information processed through Fireworks AI had been accessed by unauthorized individuals.

The exposed material included AI chat transcripts associated with Notion’s Insights feature, which helps organizations understand how their employees use artificial intelligence tools. According to Notion, fewer than 1% of its workspaces were affected.

The company clarified that most of the exposed conversations did not contain identifiers revealing the authors or the workspaces from which they originated. The dataset also excluded results generated through tool calls, limiting the additional information available to attackers.

Notion stressed that its own infrastructure was not directly compromised and that other platform functionalities were not affected.

Rapid Response Following the Security Alert

The timeline of the incident reveals how the initial assessment evolved as additional information became available.

Fireworks AI first informed Notion on November 2 that a limited dataset had been accessed. At that stage, the material was believed to consist of synthetic information created for testing purposes.

Notion subsequently disconnected its systems from Fireworks AI, removed the associated data from the provider’s environment and suspended communications between the two services.

However, further information supplied by Fireworks AI on November 6 and November 8 indicated that the exposure also involved genuine customer-related information, including AI conversation records.

Notion then reviewed the newly identified dataset, confirmed the presence of customer transcripts and notified the organizations it identified as potentially affected.

Third-Party Providers Become a Major Cybersecurity Concern

The incident illustrates a broader vulnerability within the artificial intelligence industry: companies increasingly rely on external infrastructure providers to process information, operate models and deliver AI-powered services.

Fireworks AI serves thousands of customers worldwide, including businesses developing advanced AI applications. Its infrastructure enables developers to deploy and operate open-weight AI models without having to build their own computing systems.

However, this interconnected model creates additional exposure points. A security failure at one provider can potentially affect multiple businesses and, indirectly, their customers.

The Fireworks AI incident therefore raises important questions about how technology companies monitor external suppliers, control access to sensitive information and enforce data-retention commitments.

Stronger Data Protection Measures Needed

Notion has advised affected customers to consider the exposed information compromised. Users who may have included passwords, API keys or other confidential credentials in their AI conversations should replace those credentials as a precaution.

The company has also suspended the affected Insights functionality while its investigation continues. It has indicated that customers whose workspaces have not been identified as affected do not currently need to take additional action.

Beyond the immediate consequences, the incident underscores the importance of stronger cybersecurity requirements for companies operating within the AI supply chain. As artificial intelligence becomes increasingly integrated into business operations, protecting customer information will depend not only on the security of individual platforms but also on the reliability of every external service connected to them.