OpenAI has revealed that its AI agent tool, Operator, may retain chats and screenshots for up to 90 days, even if users delete them manually. This policy is longer than the 30-day retention period set for ChatGPT. OpenAI explained that this approach aims to combat misuse and ensure the safety of its new technology.
The extended retention period provides time to review and address potential abuse risks, allowing the company to enhance fraud monitoring and refine its systems. Despite this, the policy raises concerns among users about data privacy and the duration of data retention.
Operator, launched in a research preview for subscribers of the $200-per-month ChatGPT Pro plan, is a general-purpose AI agent equipped with a built-in browser.
The tool automates tasks such as booking travel, making restaurant reservations, shopping online, and handling deliveries. Within its interface, users can choose from various categories like dining, shopping, and travel.
The operator uses screenshots of its browser to guide its actions, such as identifying buttons and filling out forms. However, it does not capture screenshots in sensitive scenarios, such as when passwords are required, which triggers a “take over” mode for users to regain manual control.
While the screenshot feature aids Operator in performing tasks efficiently, some users may feel uneasy about sharing details of their online activities with OpenAI, especially given the extended retention policy.
The company reassures users that data will only be accessible to authorized personnel and trusted service providers for purposes like abuse investigation and legal compliance.
The launch of Operator highlights OpenAI’s push to integrate AI into daily tasks, but concerns about data privacy and retention could impact user trust. As AI tools like Operator evolve, balancing functionality with privacy will be critical for maintaining user confidence in these technologies.