🤖 Overview

Operator is an AI agent developed by OpenAI, first publicly announced in January 2025, designed to autonomously perform web-based tasks on behalf of users — such as filling forms, booking travel, or ordering groceries — by controlling a remote browser instance. Unlike traditional crawlers that passively index content, Operator actively interacts with web pages as a virtual user, making it an automated legitimate agent rather than a crawler. Official documentation at platform.openai.com describes Operator as a research preview that uses a dedicated browser environment hosted by OpenAI.

🌐 Technical Behavior

Operator operates by launching a headless Chromium browser instance through OpenAI’s infrastructure, navigating to URLs specified by user prompts. It executes JavaScript, submits forms, and handles pagination, mimicking human browsing patterns at moderate speeds — typically one action every 1-3 seconds. According to OpenAI’s published IP ranges (available via their IP list at openai.com/ip-ranges), Operator traffic originates from the same blocks used by the ChatGPT service, such as 40.84.0.0/16 and 20.0.0.0/8 for IPv4, and uses HTTP/2 with TLS 1.3. The agent sends a unique User-Agent header: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36 OpenAI-Operator/1.0, and includes an X-OpenAI-Application: operator header for identification. Requests are single‑session, meaning each user task uses a fresh browser context.

📋 robots.txt Compliance

Operator fully honors robots.txt directives, as confirmed in OpenAI’s documentation. Specifically, it respects the Disallow rules for requested URLs — if a page is disallowed for GPTBot or other OpenAI crawlers, Operator will not attempt to access it. Additionally, operators can block Operator by adding User-agent: OpenAI-Operator to their robots.txt file. This compliance was verified in a March 2025 update to OpenAI’s AI safety guidelines.

🔍 Detection Indicators

The primary detection indicators are the User-Agent string containing OpenAI-Operator/1.0 and the X-OpenAI-Application: operator header. The bot also sends a Referer header pointing to the originating user’s task URL. Behavioral fingerprints include a fixed viewport size of 1280x720, consistent mouse‑event sequences (no human jitter), and the absence of browser extensions or local storage. OpenAI provides a public detection guide at platform.openai.com/docs/operator.

📊 Data Usage

Data collected by Operator — such as page content, form responses, and user interaction logs — is used solely to complete the immediate user task and to improve the underlying model’s instruction‑following abilities. OpenAI states that no personal data from the target websites is stored beyond the session duration unless the user explicitly saves it. Aggregated anonymized data may be used for model training, as detailed in OpenAI’s privacy policy (privacy.openai.com).

⚙️ Rate Limiting Policy

Operator is rate‑limited because its interactive nature can generate multiple rapid requests during a single task, potentially overwhelming poorly‑defended sites. Rate limiting is recommended at 20 requests per minute per session, with a threshold‑based block triggered after 50 requests in 60 seconds, to protect service availability without blocking legitimate automated assistance.

ⓘ Data Notice: The information presented above has been compiled from publicly available internet sources. Boteraser aggregates this data solely for informational purposes and does not independently classify, evaluate, or endorse any findings about the bots listed. The accuracy and completeness of this information is the sole responsibility of the original publishers. Boteraser and its operators accept no liability for any decisions made based on this data.