• NanoBits
  • Posts
  • Your Browser Just Got a Brain - OpenAI takes over your digital chores! 🤖

Your Browser Just Got a Brain - OpenAI takes over your digital chores! 🤖

Plus 9 OpenAI launches rewriting the rules of AI

EDITOR’S NOTE

Dear AI Explorers,

Imagine this: You're filling out yet another online form, clicking through endless dropdowns, when your screen freezes. You try refreshing, but now you can't find that specific section. Twenty minutes lost, and your coffee's gone cold.

Now imagine having an AI handle all of this for you—not in the distant future, but right now.

It turns out that last month, OpenAI released Operator, a web assistant so intuitive that it decided your entire digital life needed a “hard reboot.” Whoops!

While everyone buzzed about Operator, OpenAI quietly dropped a series of other equally fascinating launches.

From partnering with U.S. National Labs (yes, the ones where groundbreaking scientific discoveries happen) to giving European users more control over their data, they're painting a picture of AI that's both more capable and more considerate.

I'm typing this while my four browser windows are open (old habits die hard), but I can't help wondering if, in a few months, I will be dictating this newsletter to the Operator while it handles everything else.

So, grab your beverage of choice (mine’s 70% caffeine, 30% hope), and let’s dissect how OpenAI is quietly—or not so quietly—rewriting the rules of human-machine teamwork.

Stay Curious,
Yours truly!

P.S.: My browser history is a mess of half-filled forms, unbooked tickets, and abandoned online carts. Operator, if you're reading this, I could use some help!

In Today’s Newsletter:

1️⃣ The Rise of Operator: OpenAI's New Web Assistant

2️⃣ New Product & Feature Launches: o3-mini, deep research, custom GPT

3️⃣ Strategic Research: Stargate Project, Partnership with CSU

4️⃣ Content Partnerships with Axios, Schibsted Media, and The Guardian

5️⃣ European Data Residency Program

Operator by OpenAI

What is the Operator?

OpenAI's new product, Operator, is an AI that can independently complete web tasks. From booking restaurants to online shopping, it handles digital tasks without human intervention.

The impact could go beyond simple convenience. Operators might help people who struggle with technology complete essential online tasks like booking tickets, and voice command features could make the Internet more accessible for users with visual disabilities.

Other companies have similar tools: Anthropic has computer control features, and Google is developing Project Mariner. The key difference is that Operator accepts plain-language instructions, while Anthropic's tools currently require coding knowledge.

This section covers the Operator's basics, the underlying CUA technology, practical applications, current limitations, and how it relates to other AI tools in the market.

What is a Computer-Using Agent (CUA)?

Consider CUA, the tech that gives the Operator its "brain" and "hands" to use a computer like a human.

What it is: CUA is the core technology that powers Operator, OpenAI's AI agent, designed to perform tasks on the web. Think of it as the engine that drives the Operator's actions.

How it works: CUA combines GPT-4o's vision capabilities with advanced reasoning through reinforcement learning to interact with what you see on a screen.

  • Perception: First, CUA "sees" the screen by processing screenshots. It identifies important elements like buttons, text fields, and menus.

  • Reasoning: Then, it plans what to do using a "chain-of-thought reasoning" method. It breaks down tasks into smaller, manageable steps and determines how to respond to different situations, such as pop-up ads.

  • Action: Finally, CUA uses virtual mouse and keyboard inputs to click, type, scroll, and submit forms. It interacts with websites and other interfaces like a person would.

Example: Imagine you ask the Operator to book a hotel room online. CUA will:

Made with Napkin AI

Use cases of the Operator

Shopping & Delivery Services: Handles online shopping across multiple sites, grocery ordering, food delivery, and inventory management for small businesses.

Travel & Reservations: Books flights, hotels, restaurant tables, campsites, and event tickets in one place.

Business & Customer Service: Manages online orders, collects customer feedback, processes forms, and creates personalized customer experiences.

Administrative Tasks: Automates form filling, handles repetitive browser tasks, and manages digital paperwork.

Research & Information: Finds recipes, gathers information from multiple sources, and answers questions during tasks.

Accessibility & Support: Helps people with limited computer skills, assists with healthcare forms, and supports non-profits in navigating online systems.

Development & Learning: Tests AI models, builds web apps, and creates customized experiences based on user preferences.

Here is an interesting use case for using Operator with tools like Replit Agent to develop and deploy web applications.

How to use the Operator?

To use OpenAI's Operator, you can describe the task you’d like done, and the Operator can handle the rest.

Here's a breakdown of how to get started and personalize your experience:

  • Access: Operator is available to Pro users in the U.S. at operator.chatgpt.com.

  • Initiating a Task: Describe the task you want the Operator to perform. The operator is designed to work with plain-language instructions.

  • Task Execution: The operator processes and executes your instructions by interacting with websites like a human. It uses a virtual browser to navigate, click buttons, type in fields, and scroll through content. You can watch it perform tasks in real-time.

  • Monitoring Progress: You can follow the Operator's progress on the screen. The operator is powered by text-based, chain-of-thought reasoning, so it makes plans and clarifies how things can be done.

  • Clarification and Confirmation: The operator may ask clarifying questions or request confirmation for critical actions.

  • Taking Control: Users can choose to take control of the remote browser at any point. The operator is trained to proactively ask users to take over tasks requiring login, payment details, or solving CAPTCHAs. In these situations, the Operator gives control to double-check or verify information.

  • Custom Instructions: You can personalize the Operator by adding custom instructions for all sites or specific ones. This is useful for setting preferences. To add custom instructions, click on your account and select the "Websites" tab.

  • Saving Prompts: The operator lets you save prompts for quick access on the homepage, which is ideal for repeated tasks. To save a task, click the "save task" button on the top right of every conversation or task.

  • Running Multiple Tasks: By creating new conversations, you can have the operator run multiple tasks simultaneously, similar to using multiple tabs on a browser.

Safety & Privacy Aspects of the Operator

OpenAI built the Operator with safety and privacy in mind. Several key features help users stay in control.

When entering private details like passwords or payment information, the Operator switches to "takeover mode." It steps back and lets you input the information directly. During this time, it doesn't collect or screenshot anything you type. For extra-sensitive websites like email or banking, the Operator uses "watch mode," where you check each action it takes.

The operator asks for your approval before making major moves like placing orders or sending emails. It won't handle certain tasks like banking transactions or job application decisions. A special monitoring system tracks unusual behavior and can stop tasks if something seems wrong.

Through ChatGPT settings, you can opt out of sharing your data for model training. The privacy settings also let you clear all browsing data, log out of websites with one click, and easily delete your past Operator conversations.

OpenAI keeps improving these safeguards based on real user experiences, though, like any technology, it's not perfect.

Limitations of the Operator

Operator is still new and learning, and you should be aware of several limitations. It struggles with complex tasks like making slideshows or managing calendars and can get stuck when websites have tricky layouts.

When you encounter a CAPTCHA, you'll need to step in—Operator won't click those "I'm human" buttons. Many websites block AI tools like Operator, and OpenAI itself restricts access to some sites, such as Figma, Reddit, or YouTube, for technical or legal reasons.

Instead of using your browser, the Operator runs in OpenAI's data centers. You can watch and control its actions, but its web skills aren't perfect. It sometimes makes basic mistakes, misses content on pages, or enters wrong numbers. It's not great at research tasks—it works best with simple, repeated actions.

Remember that the Operator uses its own browser in OpenAI's systems, not yours. This means it can't access your personal bookmarks or settings. The tool works well for basic tasks but needs your help for anything requiring careful attention or judgment.

How to access the Operator?

The operator is currently available as a research preview.

Operator was initially available to Pro users in the U.S. However, it is now being rolled out to Pro users in other countries, including Australia, Brazil, Canada, India, Japan, Singapore, South Korea, the UK, and most places where ChatGPT is available.

You need a separate Pro subscription to use Operator through operator.chatgpt.com, which costs a whopping $200!

Operator Alternatives:

There is another AI web browsing agent, convergence.ai that does the same work for $20/month.

There's also an OpenOperator, which is 100% free and open source.

Proxy is an alternative to Operator with a free start tier.

Here's a section on the competition among AI agents, including a table comparing OpenAI's Operator with alternatives:

Competition of AI Agents

OpenAI's Operator is part of a growing field of AI agents designed to automate tasks and interact with the digital world.

Several other players, such as OpenOperator and Convergence AI, are developing similar technologies, each with strengths and weaknesses. 

Key competitors include Anthropic and Google. Other players, like DeepSeek or Meta, are anticipated to join the competition.

Here's a comparison of these AI agents in a table format:

OpenAI plans to offer this tool to Plus, Team, and Enterprise users later, making it part of the main ChatGPT experience. However, European users might have to wait longer due to local regulations.

Source: OpenAI

What’s next in the Operator’s Roadmap?

OpenAI has big plans for the Operator's future. Soon, they'll release CUA—the technology that powers Operator—through their API. This means software developers can build their own AI tools to control computers, opening up new possibilities for automation.

The Operator team is working to make the tool better at handling longer, more involved tasks. They want it to tackle complex jobs that take multiple steps and different types of work.

Currently, Operator is limited to Pro users, but OpenAI wants to make it available to everyone with a Plus, Team, or Enterprise subscription. The goal is to make the Operator part of the main ChatGPT experience. You can use it right in your chats, whether working on something immediately or setting up tasks for later.

But OpenAI is taking it slow. They want to ensure the Operator works well and stays safe as more people use it. Therefore, each new feature and expansion undergoes careful testing before being released to users.

More Updates from OpenAI

Product/Feature Launches

OpenAI o3-mini is the newest, most cost-efficient model in OpenAI's reasoning series. It is available in ChatGPT and the API. 

Why is it relevant?
It delivers STEM capabilities, with strengths in science, math, and coding, while maintaining low cost and reduced latency

Deep Research: This ChatGPT capability conducts multi-step research on the internet for complex tasks. It can accomplish in tens of minutes what would take a human many hours.

Why is it relevant?
It is useful for people who do intensive knowledge work in areas like finance, science, policy, and engineering and need thorough, precise, and reliable research. Deep research independently discovers, reasons about, and consolidates insights from across the web.

The model powering deep research scores a new high at 26.6% accuracy on Humanity’s Last Exam, an evaluation that tests AI across a broad range of subjects on expert-level questions.

More Personalizations in Custom GPT: With updated settings, OpenAI made it easier to customize the ChatGPT experience.

Why is it relevant?
Users can tell ChatGPT what traits they want it to have, how they want it to talk to them, and what rules they want it to follow. One user’s customized version of ChatGPT may respond very differently than another. 

Strategic Initiatives

The Stargate Project: This new company intends to invest $500 billion over the next four years to build new AI infrastructure for OpenAI in the United States, with $100 billion to be deployed immediately.

Why is it relevant?
The project aims to secure American leadership in AI, create jobs, generate economic benefits, support re-industrialization, and provide a strategic capability to protect national security.

SoftBank, OpenAI, Oracle, and MGX are the initial equity funders, with SoftBank and OpenAI as lead partners. Arm, Microsoft, NVIDIA, and Oracle are key initial technology partners.

OpenAI and the CSU system: The California State University (CSU) system provides more than 460,000 students and over 63,000 staff and faculty with access to ChatGPT Edu, a version of ChatGPT customized for educational institutions. 

Why is it relevant?
It is the largest implementation of ChatGPT by any single organization. This initiative will allow CSU students to integrate AI into their studies, and faculty can use it to streamline administrative tasks. CSU will also introduce a platform offering free training programs and certifications for all students, faculty, and staff.

Content Partnerships

In 2025, OpenAI formed three key media partnerships.

  • First, it is helping Axios build AI-powered newsrooms in four cities. 

  • Second, it brings news from major Nordic papers like VG and Aftenposten into ChatGPT with Schibsted Media Group.

  • Third, the Guardian partnership lets ChatGPT users read summaries of their reporting with clear links back to original articles. The Guardian is rolling out ChatGPT Enterprise across its business.

Why is it relevant?
These partnerships give ChatGPT users access to trusted news with clear source links. For publishers, they're a new way to reach readers and use AI in their work. For OpenAI, these deals help ChatGPT stay current with world events.

So far, OpenAI has joined forces with about 20 media companies, connecting their tech to more than 160 news outlets. Their content comes from hundreds of brands in over 20 languages, covering many topics across different countries. This network helps support quality journalism while making ChatGPT more useful and reliable for news-related queries.

Data Residency Program in Europe

European organizations using OpenAI's tools can now keep their data closer to home.

The new data residency program offers two key options: API customers can process their data in Europe, and new ChatGPT Enterprise and Edu customers can store their content in European data centers.

Why is this relevant?
This move responds to Europe's strict data rules. Many organizations in the EU need to know exactly where their data lives and how it's handled. By offering local data storage and processing, OpenAI makes it easier for these organizations to use AI while staying within their legal and regulatory bounds.

The program works with OpenAI's existing privacy and security measures. It's not just about where data sits – it's about giving European organizations more control over their information. Banks, healthcare providers, government agencies, and other organizations handling sensitive data can now use OpenAI's technology with greater confidence.

This expansion shows OpenAI adapting to different regional needs. As AI becomes more central to business operations, options for handling local data will likely become increasingly important for organizations worldwide.

Closing Thoughts

As OpenAI rolls out these updates in 2025, we see a clear strategy is emerging.

With Operator, they're making AI more accessible for everyday tasks. Their partnerships with news organizations and national laboratories show their commitment to information quality and scientific advancement. The European data residency program reflects their attention to regional needs and regulations.

These moves point to a future where AI is not only powerful but also practical, trustworthy, and respectful of local requirements. OpenAI seems focused on integrating AI into our daily lives while building the frameworks for responsible growth.

Until next time, keep exploring the future of AI with us.

Share the love ❤️ Tell your friends!

If you liked our newsletter, share this link with your friends and request them to subscribe too.

Check out our website to get the latest updates on AI.

Reply

or to participate.