333
Cloudflare blocking AI crawlers (lemmy.relayeasy.com)
submitted 1 year ago by to c/selfhosted@lemmy.world
33 comments fedilink hide all child comments
 

Cloudflare trying to stop AI crawling somehow!

https://arstechnica.com/tech-policy/2025/07/pay-up-or-stop-scraping-cloudflare-program-charges-bots-for-each-crawl/

you are viewing a single comment's thread
view the rest of the comments
[–] 46 points 1 year ago (6 children)

All this discussion about captchas raises a question for me: if fingerprinting is so accurate and easy, that ublock, no cookies and a VPN don't help... then why the fuck do I have to keep doing captchas?

  • permalink
  • fedilink
  • source
  • hideshow 6 child comments
    •
  • [–] 52 points 1 year ago* (last edited 1 year ago) (4 children)

    Because it never was about security. You're training LLMs for free.

    I'm pretty sure some auto drive company is getting the advantage since a lot of captchas are spotting crosswalks, traffic lights, stairs, busses, mountains, motorcycles etc. Wonder if it's fucking tesla

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 4 child comments
    •
  • [–] 13 points 1 year ago (3 children)

    I’m pretty sure some auto drive company is getting the advantage

    I'd recon that a lot of that is spliced from pictures captured from Google Map vehicles.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 3 child comments
    •
  • [–] 13 points 1 year ago* (2 children)

    Both you and @DoucheBagMcSwag@lemmy.dbzer0.com are correct. Google bought reCAPTCHA in 2012.

    Here’s an article about it from 2018.

    (╯°□°）╯︵ ┻━┻

    Captcha if you can: how you’ve been training AI for years without realising it

    And another from 2019! Captchas got harder for us because the AI had learned from our training.

    Why CAPTCHAs have gotten so difficult

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 3 points 1 year ago

    A few years ago I picked up an online gig with a company that trained AI. You'd log in to your dashboard and be presented with questions you had to answer in the best way, such as 'Is the earth round?'. Well, it's round in nature but is not perfectly round. So you'd have to pick the best solution from the answer list. It was interesting, but tedious. It put taters on the table, so I got that going for me....which is nice.

  • permalink
  • fedilink
  • source
  • parent