OpenAI’s Operator agent automates web browsing

The news: OpenAI launched Operator, an AI agent that can control a web browser and automate tasks.

• Operator is a computer-using agent (CUA) that can handle multistep tasks such as filling out forms, submitting online orders, and booking concert tickets or restaurant reservations.
• The agent is available to US ChatGPT Pro subscribers as a research preview. OpenAI plans to make it available to Plus, Team, and Enterprise users.
• OpenAI is collaborating with DoorDash, OpenTable, Instacart, and StubHub on Operator.

How it works:

• Users type requests into Operator and the AI agent uses its own web browser to view and interact with webpages like a human would.
• Operator takes screenshots to monitor progress, then analyzes those webpage images with GPT-4.
• If stuck or prompted for sensitive info, such as login credentials, it asks the user to take control. OpenAI said it should ask for permission before sending emails or making purchases.

Operator saves screenshots and chat records for 90 days, compared with ChatGPT’s 30-day retention policy, so OpenAI can review any potential abuse of the tool.

A step up: The tool’s release follows tech’s shifting focus from generative AI (genAI) chatbots toward fully autonomous bots that can act on users’ behalf.

• Anthropic launched CUA capabilities for its Claude 3.5 model in October.
• Microsoft and Salesforce are producing enterprise AI agents that can automate sales, operations, and customer service.

Rising needs: As AI models become more advanced, they can require more resources and become costlier to operate. OpenAI is already losing money on its ChatGPT Pro subscriptions due to higher-than-expected usage.

Powering systems like Operator could be a key objective for Stargate, a $500 billion AI infrastructure project to build up to 20 data centers in the US over the next four years.

OpenAI is operating Stargate with support from Arm, Microsoft, and Nvidia. Operator could transform productivity for those partners’ enterprise clients.

Our take: Operator could offer a marked jump in personal productivity and limit the need for human assistants. However, the amount of data it gathers while executing tasks could expose data vulnerabilities, and computing costs could skyrocket for OpenAI if consumers use it for every task.

This article is part of EMARKETER’s client-only subscription Briefings—daily newsletters authored by industry analysts who are experts in marketing, advertising, media, and tech trends. To help you start 2025 off on the right foot, articles like this one—delivering the latest news and insights—are completely free through January 31, 2025. If you want to learn how to get insights like these delivered to your inbox every day, and get access to our data-driven forecasts, reports, and industry benchmarks, schedule a demo with our sales team.