Crawlers, slop, and the fight for human-only spaces
A site operator said Roblox pulled 145GB of models from them, including adult content, without asking. A commenter said unless you have a real business need, you can block inbound traffic from AWS, DigitalOcean and similar ranges. Wikipedia's AI-content cleanup is on the front page, with a reference to 'low-background steel' as a metaphor for uncontaminated content. A separate Ask HN proposes a flag for AI-generated articles, and a commenter cites HN's guideline: 'Don't post generated text or AI-edited text. HN is for conversation between humans.' The 'Nietzschian Debugger' is billed as 'anti-AI' and was built with Claude, which drew mockery. Even a photo of a printer got an 'AI slop' accusation.
The pattern: scraping pressure comes in on one side and generated junk on the other, and every community is building its own defenses. Debian is the sharpest case. A maintainer's post, 'I will no longer upload to NEW', mourns that Debian is integrating an LLM into its process, and commenters say there's no alternative to Debian for them.
The counterpoint is detection. Commenters point out that AI checkers are unreliable and that people falsely accused of using AI feel unwelcome. Some say the only way to fight spam from generative AI is to use LLMs against it.
So what?
If you run a community, a data-rich site, or an open project, you need a written AI policy and bot defenses now. If you build agents or scrapers, expect blocklists and hostility. Provenance and 'verified human' features look like product gaps worth filling.
Read these
Roblox downloaded 145GB models from us (including adult content) without asking
Wikipedia's Cleanup of AI-Generated Content
Ask HN: Add flag for AI-generated articles
Nietzschian Debugger
I will no longer upload to NEW
A Printer Printing onto Paper