On July 24, 2026, OpenAI rolled out a major update for ChatGPT Work, empowering AI agents with the capability to log in to websites that require authentication. This advancement allows the AI to carry out subsequent web-based operations seamlessly, once the user has completed the initial login process. In the past, whenever ChatGPT encountered authentication barriers—such as the need for usernames, passwords, or verification codes—it was unable to proceed with the task at hand, necessitating human intervention.

With this latest update, users are only required to manually log in once within the cloud browser. Subsequently, the AI can take over and perform tasks across different sessions while maintaining an authorized status. This includes actions like downloading daily business reports, regularly monitoring for new orders, and compiling summaries for the team. The seamless continuation of tasks is made possible through session persistence technology, which ensures the login status remains active over an extended period. Importantly, this approach does not involve directly exposing user credentials to the AI, thereby upholding robust security standards.

For enterprise teams, this new feature offers a significant reduction in the engineering resources typically needed to automate business processes. However, it's worth noting that the execution of task logic will depend on OpenAI's cloud browser environment.