I get sent reports from a number of people -- mostly guys who are moderating -- that help define the needs of my anti-spam apps. Here are a couple of examples:
Blue account. I've seen all the way up to dark purple accounts:
View attachment 105127
Bot passing the CAPTCHA. This was using the same CAPTCHA style as antispambots:
View attachment 105128
Some apps with anti-spam will let users bypass the CAPTCHA if they tip, so if they're acquiring accounts with tokens they can use 1 token to allow them to push their spam.
The current series of spambots is using combining characters to make their messages harder to decipher, e.g., S͒n͒n͒a͒p͒c͒h͒a͒t͒. So anti-spam apps need to be layered and flexible to meet these different challenges.