ShieldFont is very clever, using a custom font to visually change what humans see vs. what bots see. But I strongly feel we shouldn’t make the web worse to combat AI. This font will break accessibility, search, and all sorts of other things.

@manton Interesting! I fed Claude two PDFs set in ShieldFont. It saw that something was off. In the first case, it managed to identify the known original document and corrected it. In the other, a NYT article from today, it did output garbage.

@manton could’ve just skipped the whole font trick and just force proof of work. I guess the intention is to poison AI scraping?

@manton I thought the font addressed accessibility, too.

@rom They say it does, but it sounds convoluted and I expect it will break in some readers.

@manton It shouldn’t take long for a frontier model to find a way around this. After all, AI is made of people, and people are smart.

@renevanbelzen Yep. I think these kind of hacks are short-term thinking. They make the web worse and are not going to work forever anyway.