Hot take: if I wanted to gather data via the internet, and I’m writing scripts/code to speed up the process, I have to follow some basic rules (ie look at the sitemap, find relevant robots.txt, follow that websites preference and rules).
But it seems any AI-agent I’ve used does not give af about rules and limits, and is totally cool building me a scraper that will perform hundreds of thousands of requests without regards to the website owner’s preference. Given it’s widely known you can use AI for simple coding tasks I can easily see a future where ordinary individuals are operating their own scrapers. Especially in gathering high-value information that “seems easy to get” like google search rankings, or job data.
This creates an obvious nightmare for Google, ATS platforms, and just about every website on the internet if everyone and their mother starts spinning up Playwright sessions in Python.
I’m deadset on this being a responsbility of AI providers (anthropic, open ai, anysphere, etc). But how are these companies supposed to balance this without implementing guardrails that heavily limit their products? Maybe this has been solved and someone can feed my curiosity.
[link] [comments]