**OpenAI’s Operator AI Agent Keeps Deleted Data Longer Than ChatGPT**

OpenAI’s new AI agent, **Operator**, brings enhanced functionality, but it also features lengthened data retention policies. As stated on Operator’s help page, the platform retains deleted user data — such as chats, browsing history, and screenshots — for up to **90 days**. This represents a significant extension compared to OpenAI’s ChatGPT, which keeps deleted data for just **30 days**, as initially reported by *TechCrunch*.

### Operator’s Data Retention Policy

Data retention policies are commonplace across technology platforms, typically varying by data type and intended use. Nevertheless, Operator’s functionalities, including web browsing, account logins (under user supervision), and frequent screenshot capturing to complete tasks, provide access to highly sensitive and personal data.

By default, this information is kept on OpenAI’s servers until users decide to delete it through their account settings. Even in such cases, the help page specifies that “deleted chats and related screenshots will be removed from our systems within 90 days.”

### Why Does Operator Retain Data Longer?

The prolonged retention period prompts inquiries into the disparity between Operator’s policies and those of ChatGPT. An OpenAI representative informed *Mashable* that the extended duration is designed to help the company “better understand and assess potential abuse vectors,” considering agents like Operator represent relatively new technology. This additional time assists OpenAI in refining its security protocols and curbing platform misuse.

### How Is Operator Data Utilized?

OpenAI, along with its “authorized service providers,” has access to Operator content, analogous to ChatGPT’s data policies. However, Operator’s capability to take screenshots adds a new dimension of potential surveillance. OpenAI asserts that this access is essential for investigating illegal activities, addressing misuse, providing technical assistance, or managing legal issues.

Moreover, unless users opt out, OpenAI may leverage Operator data to train its AI models. The same opt-out option available for ChatGPT is also applicable to Operator. Users who have already turned off model sharing in ChatGPT will have their Operator data excluded from training as well. To modify this setting, visit your ChatGPT account page, navigate to **Data Controls**, and switch off the “Improve the model for everyone” option.

### Security Measures for Sensitive Tasks

Acknowledging the sensitive nature of Operator’s functions, OpenAI has enacted various security measures. For example, when Operator comes across a login page, it pauses and switches to “take over mode,” returning control to the user and stopping screenshot activity. Similarly, when browsing certain sites, such as Gmail, Operator enters “watch mode,” necessitating user oversight to continue.

### Final Thoughts

Although Operator’s sophisticated capabilities render it a powerful asset for performing tasks on behalf of users, its extended data retention and screenshot features invoke significant privacy concerns. OpenAI’s rationale for these policies — enhancing security and deterring abuse — underscores the difficulties in balancing innovation with user privacy. For those worried about data usage, OpenAI offers tools to control and restrict how your information is utilized, ensuring better management of your digital footprint.