Here are some preventative measures to detect:
A bot can use multiple user accounts, so its activity is less likely to be detected.
The bot can use proxies to hide the discovery even further, obscuring its real IP address.
The bot can be programmed taking into account network rules and simply cannot break these rules in its logic.
In addition, in response to your first problem, fewer people than you might expect will actually report the problem.
In addition, and this is purely speculative , depending on network rules, one could simulate sufficient legitimate activity between two or more bots (and several user accounts) to compensate for the actual reports that are made.
In response to issue number two, with multiple accounts, the bot will simply move on to the next account when a failure occurs.
source share