
Yudkowsky advocated a policy against AI capabilities research backed by government violence (as all government policies are). However, some people say that this will get misinterpreted as a call to individual violence/terrorism.
This question doesn't have exact resolution criteria because for all I know terrorist attacks could look pretty crazy. However here's a rough anchor that it will be compared against:
There should be actual death or intent to kill. Material sabotage is interesting in its own right but not what this question is about.
There should be independent evidence that the person in question is a fan of Eliezer or otherwise directly influenced mainly by Eliezer's writings, but this evidence is pretty broad and includes e.g. commenting on LessWrong or using a lot of rationalist lingo.
There should not be evidence that the terrorist understood the limits of what Yudkowsky was advocating for. For example if there is a manifesto where the terrorist complains about Yudkowsky not going far enough due to opposing violence, then that would disqualify them.