[teraflop]

By that reasoning, a lock has the capacity to grant or deny access to whatever is behind a locked door. And if I pick the lock, well, that just means I was sufficiently persuasive that the lock agreed to let me in, doesn't it? Clearly, by using a lock that opens in response to certain inputs, the owner is choosing to grant access to anybody who provides those inputs.

I'm not trying to argue that guessing sequential IDs in a URL is morally the same as picking a lock. I'm arguing that in both cases, there's no human in the loop, so it's not at all obvious to what extent a human should be assigned responsibility. In your example, the letter does not have agency, but the CEO certainly does; and if weev had written 110,000 letters to AT&T that were read and responded to by humans, I can't imagine how there would be any case against him.

See also: the debates surrounding Google's autonomous cars, or the Do-Not-Track header.