Virtually any ISP that is blocking mail locally is also smart enough to figure out when a sender-address is being forged. In most cases, the IP of a message's origin is what ISPs block on, not the sender's domain. In extreme cases (where all mail from a domain is known to be rogue, or when a domain is known to never send mail from that domain), a domain, or hosts within a particular domain, might be blocked, but this is fairly rare.
AOL doesn't reject messages outright a lot of the time (this is slowly changing, supposedly). When a message is rejected during the SMTP transaction itself, the job of delivering the bounce falls upon the sending MTA. Viruses and spamware that send direct to the MX generally ignore the failure completely, so the only time you'll get a bounce is when the message is accepted by a mail server and then rejected to the sender.
It's understandable why AOL does this - they have a lot of users, so their mail system is more complicated, and rejecting unknown addresses immediately would make it easier for spammers to "dictionary attack" their mail servers to find valid usernames. However, the sheer volume of bounces can overwhelm a mail server. If it's a long term, persistent problem, they can often be convinced to disable bounces for a particular sender-address or domain.