- Barracuda Research reveals the scope of data scraping bots
- Not all bots are bad, but many extract large amounts of data without permission
- These “gray bots” can be very aggressive, he reports
A new investigation by Barracuda has identified “Bots Grises”, along with good and bad bots that track the web and extract data, and although the “good bots”, such as SEO and customer service bots, seek information, the “BAD Bots” are designed for harmful activities such as fraud, data data and breach accounts.
In the space between, there are “Gray Bots”, which Baraccuda explains are Genai scraper bots designed to extract serious amounts of data from the websites, most likely to train AI models, or to collect web content such as news, reviews and travel offers.
These bots are “blurring the limits of legitimate activity,” argues the report. While they are not absolutely malicious, their approach can be “questionable” and some are even “very aggressive.”
Major activity
Baraccuda’s detection software found millions of applications received by Genai Bots web applications between December 2024 and February 2025, with a followed web application that received 9.7 million requests from Bot Bot in just 30 days.
These bots collect data and can delete them without permission, and can also overwhelm web applications with traffic, interrupt operations and take data protected by copyright to train AI models, which may be in violation of the rights of the owner.
There have been many setbacks against practices such as these, with creative industries in the United Kingdom that launched a campaign of ‘Make It Fair’ to protest against their work using AI models to create photos, videos, stories or other content without permission or credit.
Data privacy risks also come with this level of scraping, since some sites carry confidential data of customers, for example, those in health services or financial services.
Bots can also obscure the analysis of websites, which makes organizations difficult to evaluate and track genuine traffic or user behavior, making the most difficult commercial decisions.
