Making “anti-ai” fonts that obfuscate or scramble text is a fools errand. For fonts that scramble text, it's an accessibility problem, since the scrambled text is what screen readers and other accessibility tools will parse. Already, you've cut out many of the humans you're trying to reach.
In case you aren't aware of what I'm referring to, here are a few examples:
Edit: I'd like to preemptively clarify I don't have anything against these specific examples, nor am I trying to invalidate the work that went into creating them or any other examples of anti-AI fonts. This post is meant to critique the idea itself. I'm not trying to put-down anyone here!
There is just no way to do this accessibly, since an accessible solution will require machine-accessible metadata. This pushes us back to the giant issue of human verification, and all the privacy and security concerns associated with it. Ultimately, trying to solve the problem of selectively granting access to metadata for disabled people will lead towards filtering content through some centralized identity verification system. Unless you're a Nazi or some sort of sociopath, creating large lists of disabled people is not a road I think we want to travel down.
The public posts and discussions being had about this subject are already informing AI companies on how to train their multimodal models to get around these obfuscations, most of which have already been broken. I'd argue every new font and tech demo is effectively a benchmark, daring AI firms come up with solutions to sidestep them. And they will be sidestepped, one way or another. If a human can see the information, that means there is a way the information can be parsed. "Ghost" fonts will become just another scraping obstacle with its own set of contingencies.
Some of the more elaborate demos I've seen involving motion graphics and videos are impractical for normal use on websites and therefore are irrelevant as serious long-term solutions to this problem. Besides, if an effective obfuscation method somehow gains widespread use, then a much higher priority will be placed on breaking it, eventually rendering it useless.
A cat-and-mouse game like this is pointless if it results in a world where all websites are highly obfuscated, requiring computationally expensive systems to access the content legitimately or illegitimately. The end result will probably be some sort of copy-protection system which will benefit those who want to censor the web and create paywalls and other gatekeeping methods to block access to content. We're better off with the web as it is now, written entirely in plaintext. Besides, obfuscation goes against the spirit the World Wide Web was founded on: free and open access to information for the benefit of humanity.
I don't think there will ever be a silver bullet for this problem. While the speed at which their capabilities will increase is highly debatable, AI systems are never going to start getting dumber. Like it or not, all publicly available information will inevitably become accessible to anyone and anything that has permission to access it. This is the baseline scenario people need to plan around.