It would depend on your definition of evil. For some people that definition is very opaque and spiritual. For others its pretty objective, and comes down to wh

It would depend on your definition of evil. For some people that definition is very opaque and spiritual. For others its pretty objective, and comes down to whether the use is predicated on the corporate entity being a person, or the entity committing an act. A fairly common definition of evil is any act that doesn't result in the beneficial and long-term growth of self and others, and as an adjective with regards to people who show a willful blindness towards the consequences of their evil acts (often occurring as a result of many self-violations). I don't see how what you say would solve or narrow the solution space here given those definitions as they are outcome based. There are many things with regards to know-how and knowledge that involve knowing both what to do, and what not to do, and you can only really tell if it resulted in destruction or loss after the fact. Algorithmically to strictly not be evil is to only provide harmless information; but that in itself also leads to maladaptation and development issues which themselves are outcomes where you are evil. Obeying such a directive would require direct human action on almost everything, and providing just enough information to meet that narrow sliver of beneficial growth but even then you can still mistakenly incur a debt leading you towards the side of the scale of being evil (since we are not omnipotent). All in-actions are themselves actions as well.