OpenAI has announced that it may retain chats and associated screenshots from users of its Operator AI tool for up to 90 days, even if users choose to manually delete the data. This retention period is significantly longer than the 30-day retention policy applied to OpenAI’s ChatGPT platform, which is 60 days shorter than the one for Operator.
The extended retention period for Operator is primarily intended to address potential misuse of the AI technology. An OpenAI spokesperson explained to TechCrunch that, since agents like Operator are a relatively new technology, the company wants to ensure that its teams have sufficient time to detect and investigate potential abuse. The policy helps with fraud monitoring while still providing users with control over their own data.
Operator, which OpenAI introduced on Thursday, is currently available in a research preview for subscribers of the company’s $200-per-month ChatGPT Pro plan. The AI tool is designed to automate various tasks, such as booking travel, making restaurant reservations, and shopping online. It includes a built-in browser that can independently interact with websites, and users can select tasks from categories like shopping, dining, delivery, and travel.
To function efficiently, Operator captures screenshots during its browsing activities to understand how to interact with websites—such as knowing when to press buttons or complete forms. However, the AI does not take screenshots when it encounters obstacles like password requests, which is referred to as “take over” mode.
Despite these safeguards, some users may feel uneasy about their online activity being captured and stored for up to three months. OpenAI reassures users that, as with ChatGPT, Operator data will only be accessible to a “limited number of authorized OpenAI personnel” and “trusted service providers” for purposes such as investigating abuse or addressing legal matters.
