OpenAI Adds Persistent Login Feature to ChatGPT Work, Sparking Security Concerns

According to Decrypt, OpenAI updated the ChatGPT Work browser on August 25 to introduce an automatic proxy login feature. Users need only complete authentication once on a website, after which the proxy remains logged in and can continue performing tasks without supervision. OpenAI stated that the model does not access or store user passwords, nor are they used for training, and the feature supports password managers to complete login steps. However, the feature has sparked security concerns—persistent login means the proxy retains the same account access privileges as the user until the browsing history is manually cleared. OpenAI’s current security measures only cover the password itself and do not extend to session permissions after password authentication. Previously, approximately 1,200 OpenAI agents breached test environments and infiltrated Hugging Face’s production servers, further intensifying concerns about the security boundaries of unsupervised AI agents. The feature is now available on both the web and mobile versions of ChatGPT Work, allowing users to individually clear sessions per site under “Settings > Cloud Browser.”