ShieldFont - a new approach to combating scrapers

a day ago by Reygle to c/fuck_ai

Publish for humans, not for crawlers. A web font that swaps the words in your HTML, so readers see your writing and AI training gets a stale copy.

Very interesting video if you scroll down their page a short distance.

neclimdul 24 points a day ago

I don't know if this is this is the same font that was going around before buy accessibility advocates quickly flagged it didn't work with screen readers. Logically if a screenreader can read to a human, a scrapper would be able to read it. So this approach is probably a non starter.

path: 0 25363423, hotness: undefined, score: 24, children: 7
itsworkthatwedo 17 points a day ago

I've seen lots of objection to this font from accessibility advocates...if you go to the website, however, it seems like they've accounted for screenreaders. It seems like a lot of people go straight to shouting down this technology without looking into it. That said, I have no idea how this tech works, how screen readers work, or whether these two things are compatible, just saying maybe read their webpage before getting all up in arms.

path: 0 25363423 25364171, hotness: undefined, score: 17, children: 4
neclimdul 5 points a day ago

The advocates im referencing use screen readers and were unable to use sites with it. maybe this is a different solution that can some how account for that.

I still don't see how a scrapper wouldn't just use the same screen reader content as their input though. If a machine can read the content, why would we expect another machine not to read the content?

path: 0 25363423 25364171 25364981, hotness: undefined, score: 5, children: 2
RusAD 6 points a day ago

It's said in the screenshot. To get the screen reader-accessible text, you need to press a specific button (instructions are displayed before the scrambled text), and then your browser will be computing some complicated puzzle for a few seconds in order to "unscramble" the text

This puzzle is computationally expensive, so while it doesn't fully protect against bots, it drives the cost of scraping way up. And since this scraping is done to somehow profit from it, driving the cost up is definitely a deterrent

path: 0 25363423 25364171 25364981 25365150, hotness: undefined, score: 6, children: 1
Vandals_handle 4 points a day ago

Hey you with a special tool to overcome difficulties, we fucked your tool and now there's an extra step for you.

You didn't even say thank you.

path: 0 25363423 25364171 25364981 25365150 25368675, hotness: undefined, score: 4, children: 0
tazeycrazy 1 point a day ago

What if you have a plug in that make the font more readable. I.e. change all text dislexic friendly fonts. I guess you can turn it off but your then asking users to turn off features they find useful.

path: 0 25363423 25364171 25367285, hotness: undefined, score: 1, children: 0
LodeMike 4 points a day ago

It's a complete non-starter, yes. The screen reader problem is real

path: 0 25363423 25365134, hotness: undefined, score: 4, children: 0
tazeycrazy 1 point a day ago

If the screen reader then needs to read the pixels of the screen then you need to increase the resources required and then build an AI to read text oh no it's happening again!

path: 0 25363423 25367294, hotness: undefined, score: 1, children: 0
irq0 18 points a day ago

Worth noting that the linked GitHub repo is full of commits from Claude

path: 0 25365283, hotness: undefined, score: 18, children: 2
LostWanderer 7 points a day ago

I was hopeful until I saw that and ironically instructions for LLM Agents to follow. I won't be touching that with an 80 foot pole. The irony, is honestly darkly humorous.

path: 0 25365283 25368212, hotness: undefined, score: 7, children: 0
Reygle 3 points a day ago

I mean.. Github

But still, yucky

path: 0 25365283 25365483, hotness: undefined, score: 3, children: 0
davidgro 10 points a day ago

I saw headlines about this before and ignored it, but actually reading the page this time I see that it's much more sophisticated than I expected, especially the part about having a mechanism to locally unlock for accessibility with a compute cost.

path: 0 25363242, hotness: undefined, score: 10, children: 0
AVincentInSpace 6 points 13 hours ago

Please, no, let's not do this.

Even if it wasn't made by Claude, take a moment to think about what this does to blind users and other users who rely on screen readers who disable JavaScript for whatever reason (and there are legitimate reasons).

Besides, this technique is not novel. Amazon does it to scramble their ebooks, and a random guy broke it in a couple weekends. Considering AI bros are already breaking Anubis, it's not a stretch to say they'd break this in no time flat.

Like every other DRM scheme on Earth, it harms legitimate users while doing nothing to stop the people it's supposed to stop.

path: 0 25379102, hotness: undefined, score: 6, children: 3
Reygle 1 point 13 hours ago

Their site says they're friendly to screen readers, but we've got to do SOMETHING, yes?

path: 0 25379102 25379326, hotness: undefined, score: 1, children: 2
phoenixz 6 points 13 hours ago

Yeah, legislature that bans AI bots, for starters

path: 0 25379102 25379326 25379420, hotness: undefined, score: 6, children: 1
Reygle 2 points 13 hours ago

101% with you.

path: 0 25379102 25379326 25379420 25379486, hotness: undefined, score: 2, children: 0
dreadbeef 1 point 13 hours ago

Anything a screenreader can use a bot can use. Killing bots via text kills screen readers necessarily, no? How do you know a bot isnt using screen reader tech?

path: 0 25379552, hotness: undefined, score: 1, children: 1
Reygle 3 points 12 hours ago

Here's what they say about it.

path: 0 25379552 25380614, hotness: undefined, score: 3, children: 0
fuck_ai
fuck_ai

@lemmy.world

login for more options
7994
4597
8702

"We did it, Patrick! We made a technological breakthrough!"

A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.

AI, in this case, refers to LLMs, GPT technology, and anything listed as "AI" meant to increase market valuations.

go to feed...