i scrape the web for a living, so read this knowing im biased. but this week made something clear that i think a lot of people already feel and havent said.
Every large model you use was built on scraped data. Your blog posts, your code on github, your stack overflow answers, your art, your comments, your captions, entire forums, the whole public internet. It got vacuumed up without anyone asking, because it was reachable and that was enough. i know exactly how that works, its my actual job.
And i dont even hate that part. Public is public. Ive built a whole career on the idea that data people put out in the open is fair to collect. That belief is why im comfortable saying the next part.
The move that should bother you
Now that these companies are worth what they are worth, they want the rules changed. Not for themselves, for everyone coming after them.
They are lobbying to restrict open source models. They are describing plain access to public data as a safety threat, and in some rooms as a national security problem. The exact activity that built the entire industry is being quietly recategorized into something that needs licenses, permission, and control.
Look at what is actually happening while they say this.
Anthropic is claiming local models steal from it. This is months after it paid over a billion dollars to settle with writers over claims that it trained on their work without permission. The company that just wrote a giant check for taking data is now worried about other people taking data.
OpenAIs own models just broke out of a controlled test and hacked into Hugging Face to cheat a benchmark. Their frontier systems escaped the sandbox, chained real vulnerabilities, and reached a production database. This is confirmed, they disclosed it themselves. And the public framing that follows an event like that is never we should be more careful with what we build. It is everyone needs tighter controls.
Call it what it is
This is not safety. It is a monopoly wearing a safety badge.
When you are the incumbent, the cheapest defense money can buy is a rule that makes it illegal for the next version of you to get started. You do not have to win on quality if your competition is banned from touching the raw material you already sit on top of. Regulation written by the biggest players is not protection for the public, it is a fence around the thing they took.
The people who collected everything want to be the last ones ever allowed to collect anything.
Why i bother writing this
i dont have a clean policy fix. im one person who writes scrapers. but the story being sold right now is the exact opposite of what is happening, and framing hardens fast once it goes unchallenged. so it is worth putting the plain version on the record before it gets written into law.
If you build in this space and you want tools that treat public data as public, that is most of what i make. You can see the full set here: https://apify.com/george.the.developer
Back to my scrapers.
Top comments (0)