GPTBot

2023: The beginning of blocking by Roskomnadzor

On December 11, 2023 Roskomnadzor , he began sending letters to hosting providers with recommendations for identifying and blocking the company's GPTBot bot. OpenAI According to the department, such measures are necessary to "exclude the collection of information on critical resource vulnerabilities."

GPTBot, which appeared in August 2023, is designed to scan websites and collect open data that can subsequently be used to train and improve the ChatGPT neural network. However, after the launch of the bot, OpenAI was criticized for unauthorized data collection. At the same time, experts express concerns that AI bots can be used for cybercriminal purposes, for example, to quickly hack network resources, phishing, ensure the operation of darknet sites, etc.

Roskomnadzor began sending letters to hosting providers with recommendations for identifying and blocking the GPTBot bot

Against this background, according to the Kommersant newspaper, the Main Radio Frequency Center (GRCC), subordinate to Roskomnadzor, declares the need to assess the risks associated with the work of GPTBot: we are talking about the possible collection of information about weak points of web resources or "other sensitive information, including containing personal data." If such risks are identified, the GRCC indicates the need to block the bot's circulation.

Meanwhile, OpenAI itself has published instructions on how to restrict GPTBot access to the site or provide it with data only partially. In addition, the company publishes a list of IP addresses from which the bot is accessed. Thus, owners of Internet resources can, at their discretion, block GPTBot access to their sites.

GPTBot itself is designed as correctly as possible, it always "appears" when it appears on the network, and declares compliance with the access rules for bots set by resource owners, "emphasizes Qrator Labs specialist Georgy Tarasov.[1]

Notes