OpenAI launches cloud browser for ChatGPT Work to automate web tasks
OpenAI has introduced a cloud browser for ChatGPT Work, allowing the AI to perform web-based tasks such as navigating pages, filling forms, and interacting with signed-in websites. The feature runs on a remote computer, ensuring it does not access the user’s local browser data like passwords or cookies.
Editor, Lazyfounder

30 SEC SUMMARY
- OpenAI has added a cloud browser to ChatGPT Work, enabling it to perform web tasks like navigating pages, filling forms, and completing actions on a user’s behalf.
- The cloud browser operates on a remote computer, avoiding access to the user’s local browser data such as passwords, cookies, or extensions.
- It supports signed-in websites by allowing users to log in through a secure form, without exposing credentials to the AI model.
- OpenAI designed the feature to seek user confirmation before actions involving financial, legal, or account-related commitments.
KEY HIGHLIGHTS
- OpenAI introduces a cloud browser for ChatGPT Work, enabling it to open websites, navigate pages, and complete online tasks.
- The cloud browser runs on a remote computer, avoiding access to the user’s local browser data like passwords, cookies, or extensions.
- Users can log in to supported websites through a secure form, with credentials sent directly to the remote browser and not stored by ChatGPT.
- ChatGPT asks for user confirmation before performing actions that could create financial, legal, or real-world commitments.
How the cloud browser works
The cloud browser operates on a separate remote computer, rather than the user’s local device. According to YourStory, this setup prevents it from accessing open tabs, saved passwords, cookies, extensions, or browsing history stored on the user’s personal device.
ChatGPT Work uses this remote browser session to read webpages, click buttons, and enter details where supported. This could include tasks like checking restaurant availability, comparing product stock, tracking packages, or preparing bookings.
Accessing signed-in websites
The cloud browser can interact with signed-in websites without exposing the user’s credentials to the AI model. When ChatGPT encounters a login page, it pauses and prompts the user to sign in through a secure form.
Users enter their username, password, and two-factor authentication codes themselves. YourStory reports that these credentials are sent directly to the remote browser and are not visible to the AI model or stored by ChatGPT. Once logged in, ChatGPT can continue working within that session until it expires or the user clears the browser data.
This feature addresses a major limitation for AI agents, allowing ChatGPT to access online dashboards, travel portals, utility accounts, or workplace systems that require authentication.
Safeguards and limitations
OpenAI has designed ChatGPT to ask for user confirmation before performing actions that could create financial, legal, or real-world commitments. This means the AI can assist with preparatory tasks but leaves final decisions to the user.
The feature may still face limitations. Some websites may block automated browser agents, while others may require the user to complete the final step manually. Additionally, users are advised to avoid entering sensitive information like passwords or payment details directly into the chat. Instead, they should use the secure sign-in flow or takeover mode for such tasks.
Users can manage website permissions and clear cloud browser data as needed, providing some control over the AI’s access to their accounts.
What this means
Lazyfounder analysis — our interpretation, not reported fact.
This update signals a shift in how AI assistants could integrate into daily workflows, particularly for founders and operators who rely on web-based tools for tasks like tracking orders, managing bookings, or accessing business dashboards. By enabling ChatGPT to interact with signed-in websites, OpenAI is addressing a long-standing limitation for AI agents—accessing secure, user-specific data without compromising privacy. However, the feature’s utility will still depend on website compatibility and user oversight, especially for high-stakes actions. For startups building AI-driven tools, this move underscores the growing expectation for AI to handle more than just information retrieval—it must also execute tasks autonomously, albeit with guardrails.
Key takeaways
- Founders and operators can automate repetitive web tasks like checking restaurant availability, tracking packages, or preparing bookings, reducing manual effort for routine workflows.
- The feature’s reliance on remote browsing sessions could set a new standard for AI privacy, particularly for tools that interact with sensitive business or personal data.
- Startups developing AI agents may need to account for signed-in website compatibility, as users increasingly expect AI to navigate secure portals and dashboards.
- OpenAI’s approach—requiring user confirmation for high-stakes actions—highlights the importance of balancing automation with control, a consideration for any AI tool handling financial or legal tasks.
Sources
This story is an original summary drafted with AI by Lazyfounder from the reporting listed above and checked by automated validation. Facts are attributed to their original publishers; sections marked as analysis are Lazyfounder's. Where a source is in another language, facts were machine-translated and quotations are reported, not reproduced. Read the original coverage via the links, and see our AI policy and corrections policy.
About the author
Editor, Lazyfounder
Tarun Mottlia edits LazyFounders, covering Indian startups, funding rounds, AI and product launches. Every story on the site is AI-assisted and checked against its cited sources before publication.
More stories by Tarun MottliaGet the LazyFounder Brief
Startup, funding and AI news in a five-minute read. Join the early-access list.


