On 9 November 2025, Mixpanel, a third-party analytics service used by OpenAI for tracking usage on its API platform, detected that an attacker had gained unauthorized access to a portion of its systems and exported a dataset containing customer-identifiable and analytics information.
On 25 November 2025 Mixpanel shared the affected dataset with OpenAI for review, and on 26 November 2025 OpenAI publicly disclosed the incident.
What Data Was Exposed
According to OpenAI, the exposed information was limited and related only to "analytics-level" account metadata for some API users.
Potentially exposed data included:
- Names provided on the API account
- Email addresses linked to the API account
- Approximate location inferred from browser metadata (city state or country)
- Operating system and browser used to access the account
- Referring websites (i.e. from where the user came before logging in)
- Organisation or user IDs associated with the API account
This data did not include passwords API keys payment information or chat content.
What Was Not Exposed
OpenAI confirmed that this incident did not involve any breach of its own infrastructure.
The following remained safe and were not exposed:
- Chat content prompts or API conversation data
- Passwords authentication tokens or login credentials
- API keys or developer access secrets
- Credit card or payment details financial records billing addresses
- Sensitive identity information such as government-issued IDs
Actions Taken by OpenAI
After receiving confirmation of the exported dataset OpenAI immediately removed Mixpanel from its production services. A security investigation was started to analyse the scope of the exported data. OpenAI also notified impacted organisations administrators and account holders privately. In addition OpenAI reviewed its vendor security policies and tightened security standards for third-party analytics and service partners.
What Users Should Consider Doing
Even though the leaked data was limited and did not include credentials or content the exposed metadata could still be used in phishing or social-engineering attempts.
If you use the OpenAI API (or manage an API account):
- Be alert to unexpected or suspicious emails or messages claiming to come from OpenAI or related services.
- Verify the sender domain before responding especially if the message requests credentials API keys or personal data.
- Enable multi-factor authentication (MFA) if it is available for your account or organisation.
- Avoid sharing more personal data with external analytics tools than what is absolutely necessary.
Why This Incident Matters
This episode is a reminder that even if a company's core systems are secure using external analytics or vendor tools introduces additional risk. A breach at a third-party provider can expose user-related metadata in a way that helps attackers build convincing phishing or social-engineering campaigns. For anyone building applications or testing or securing API products this also shows why vendor security data minimisation and strict integration policies are no longer optional but critical parts of the security design.
References
- OpenAI. What to know about a recent Mixpanel security incident, 26 November 2025. https://openai.com/index/mixpanel-incident/
- DEV Community. The OpenAI Mixpanel Security Incident Explained, 26 November 2025. https://dev.to/alifar/the-openai-mixpanel-security-incident-explained-2ok5
- Moneycontrol. OpenAI Discloses Mixpanel Security Incident Says No API Data or Credentials Exposed, 26 November 2025. https://www.moneycontrol.com/artificial-intelligence/openai-discloses-mixpanel-security-incident-says-no-api-data-or-credentials-exposed-article-13698801.html
- The Cyber Express. OpenAI Discloses Mixpanel Security Incident Without Leaking API Data, 26 November 2025. https://thecyberexpress.com/openai-mixpanel-security-incident/
