It is one thing to find and neutralize a virus or malware. But it would be quite another thing if the AV (Anti-Virus) could turn around and attack the virus source. Every virus or malware serves a purpose, mostly to try and either "phone home" or "allow someone from home to phone in". This weakness leaves open a door to the virus- / hacker-author which can and should be exploited, either directly or via a proxy. If a hacker sends out a worm or trojan, would it not be nice if his servers and environment could be attacked by every machine infected? Even if it were just to result in a DoS attack on them, it would be more than they suffer now? They usually cast their nets as wide as possible, what if that can be used against the virus? What if their nets catch more than they bargained on?
Take for example a case where a Trojan JS downloader is used to install ransomware. The ransomware therefore comes from somewhere, either the hackers environment or yet another infected machine. The ransomware contains links to the payment sites, usually much closer to the hacker's home? Even if the Anti-Malware were to clean the virus and then attack an infected (and may be previously innocent machine), it would force the IT admin of said machine to take action to remedy the infected machine, for ex. taking it offline - instead of not knowing and leaving it to spread the virus.