Meta has quietly unleashed a new web crawler to scour the internet and collect data en masse to feed its AI model.
The crawler, named the Meta External Agent, was launched last month, according to three firms that track web scrapers and bots across the web. The automated bot essentially copies, or “scrapes,” all the data that is publicly displayed on websites, for example the text in news articles or the conversations in online discussion groups.
A representative of Dark Visitors, which offers a tool for website owners to automatically block all known scraper bots, said Meta External Agent is analogous to OpenAI’s GPTBot, which scrapes the web for AI training data. Two other entities involved in tracking web scrapers confirmed the bot’s existence and its use for gathering AI training data.
While close to 25% of the world’s most popular websites now block GPTBot, only 2% are blocking Meta’s new bot, data from Dark Visitors shows.
Earlier this year, Mark Zuckerberg, Meta’s cofounder and longtime CEO, boasted on an earnings call that his company’s social platforms had amassed a data set for AI training that was even “greater than the Common Crawl,” an entity that has scraped roughly 3 billion web pages each month since 2011.
I think of this as a problem with opt-in only systems. Think of how sites ask you to opt in to allow tracking cookies every goddamn time a page loads. A rule based system which lets you opt in and opt out, like robots.txt, to let you opt out of cookie requests and tell all sites to fuck would be great. @aniki@lemm.ee is complaining about malicious instances of crawlers that ignore those rules (assuming they’re right and that the rules are set up correctly), and lumping that malware with software made by established corporations. However, Meta and other big tech companies haven’t historically had a problem with ignoring configurations like robots.txt. They have had an issue with using the data they scrape in ways that are different than what they claimed they would, or scraping data from a site that does not allow scraping by coming at it via a URL on a page that it legitimately scraped, but that’s not the kind of shenanigans this article is about, as meta is being pretty upfront about what they’re doing with the data. At least after they announced it existed.
An opt-in only solution would just lead to a world where all hosts were being constantly bombarded with requests to opt in. My major take away from how meta handled this is that you should configure any site you own to disallow any action from bots you don’t recognize. As much as reddit can fuck off, I don’t disagree with their move to change their configuration to:
User-agent: * Disallow: /Removed by mod
Oh, no, that wasn’t excusing Meta in general. Just giving them a pass on that they’ve had, to my knowledge, a history of respecting robots.txt, which makes this piece of software better than outright malware. Starting it secretly and not giving site hosts a chance to make sure they had their privacy configured the way they liked first was a shady as hell move, no argument there.
Removed by mod