Notion AI Breach Exposes Customer Chat Transcripts

What happened at Fireworks AI

Notion announced that its AI infrastructure provider, Fireworks AI, suffered a security incident that allowed unauthorized access to a limited set of Notion AI chat logs. The breach was discovered during a routine security review and was reported to Notion’s security team within days of detection.

Scope of the data exposure

The incident affected a small fraction of Notion workspaces. Only chat transcripts generated by the Notion AI feature were accessed, and no other user data, such as files, passwords, or personal identifiers, was reported as compromised. The exposed logs contained the text of user queries and the AI-generated responses.

Key facts about the breach

• Source of breach: Fireworks AI’s cloud environment.
• Data accessed: Notion AI chat transcripts from a limited number of workspaces.
• Duration of exposure: The unauthorized access window lasted less than 48 hours.
• Impact on users: No evidence of further misuse or distribution of the transcripts.

How Notion responded

Notion’s security team activated its incident response plan immediately after learning of the breach. The company took the following steps:

1. Isolated the affected Fireworks AI resources to prevent additional access.
2. Conducted a forensic investigation with the help of external security consultants.
3. Notified the affected workspace owners and provided guidance on reviewing their AI chat history.
4. Implemented additional encryption controls for AI data in transit and at rest.
5. Published a detailed post on the Notion security page outlining the incident and remediation steps.

The company emphasized that the breach did not affect core Notion functionality, and regular workspace data remains protected.

Implications for users

While the breach involved a narrow data set, it highlights the broader risk of relying on third‑party services for AI processing. Users should be aware that any content entered into AI tools may be stored outside the primary platform and could be subject to separate security controls.

Potential privacy concerns

AI chat transcripts can contain sensitive business information, project details, or personal queries. Even when only a small number of logs are exposed, the content could reveal strategic plans or confidential discussions.

Regulatory perspective

Data protection regulations such as the EU General Data Protection Regulation (GDPR) require organizations to ensure that any processing by third parties meets strict security standards. Notion’s swift notification aligns with the GDPR’s breach‑notification timeline, which mandates reporting within 72 hours of discovery.

Best practices for protecting workspace data

To reduce the risk of similar incidents, users can adopt several precautionary measures:

• Review AI usage policies and limit the input of highly confidential information.
• Enable two‑factor authentication on all Notion accounts.
• Regularly audit third‑party integrations for security certifications.
• Keep software and browser extensions up to date to minimise exposure to known vulnerabilities.
• Consult the Cybersecurity and Infrastructure Security Agency guidance on cloud security best practices.

Industry reaction

Security analysts have noted that the breach underscores the growing interdependence between productivity platforms and specialised AI providers. The Verge reported that several firms are reevaluating their AI vendor contracts to include stricter data‑handling clauses.

Fireworks AI released a statement confirming that the incident was isolated, that the compromised environment has been secured, and that they are working with Notion to enhance joint security controls.

What users can do now

Workspace owners who received a breach notification should take the following actions:

1. Log into Notion and review recent AI chat history for any sensitive content.
2. Export or delete any transcripts that contain information you prefer not to retain.
3. Communicate with team members about the incident and reinforce data‑handling guidelines.
4. Consider disabling the Notion AI feature temporarily if your organization handles highly classified data.

By staying informed and applying these safeguards, users can continue to benefit from AI‑enhanced productivity while minimising exposure to future incidents.