OpenAI revealed a security incident in late- November 2025 that allowed hackers to access data about users via its third-party analytics provider, Mixpanel. The breach, which took place on November 9, 2025, exposed a small amount of personally identifiable information for some OpenAI API users, although OpenAI stressed that its own systems had not been the target of the attack.
Breach details
The breach occurred completely within Mixpanel’s own infrastructure, when an attacker was able to gain access and exfiltrate a dataset containing customer data. Mixpanel became aware of the compromise on 9 November 2025, and following an investigation, shared the breached dataset with OpenAI on 25 November, allowing the technology firm to understand the extent of potential exposure.
The breach specifically affected users who accessed OpenAI's API via platform.openai.com, rather than regular ChatGPT users. The compromised data included several categories of user information collected through Mixpanel's analytics platform. Names provided to accounts on platform.openai.com were exposed, along with email addresses linked to API accounts.
Additionally, coarse approximate location data determined by IP addresses, operating system and browser types, referring websites, and organization and user IDs saved in API accounts were part of the breach. However, OpenAI confirmed that more sensitive information remained secure, including chat content, API requests, API usage data, passwords, credentials, API keys, payment details, and government IDs.
Following the incident, OpenAI took immediate action by removing Mixpanel from its services while conducting its investigation. The company notified affected users on November 26, 2025, right before Thanksgiving, providing details about the breach and emphasizing that it was not a compromise of OpenAI's own systems. OpenAI has suspended its integration with Mixpanel pending a thorough investigation of the incident.
Recommended measures
OpenAI also encouraged the affected users to stay on guard for potential second wave attacks using the stolen information. Users need to be especially vigilant for phishing and social engineer attacks that could be facilitated by the leaked information, such as names, e-mail addresses and company information. A class action has also been brought against OpenAI and Mixpanel, claiming the companies did nothing to stop the breach of data that revealed personally identifiable information for thousands of users.