So far the only thing I saw that could break previous "squiggly text" style captchas was some research papers. I have yet to see any of it used in the wild. So

So far the only thing I saw that could break previous "squiggly text" style captchas was some research papers. I have yet to see any of it used in the wild. So I doubt spammers brought this upon us - the previous captchas were more than effective at keeping them at bay already. The real reason is shitty Google being greedy and needing someone to label their training data.