Chemical elements and compounds have nature, species have behavior. “Human nature” is a beautiful literary term, but unfortunately it is useless in our empirical world. In the empirical world killing, stealing and enslaving is not human nature – it is human behavior. And human behavior is not rooted exclusively in human genes, it is also a subject to environmental influences.
Even if we could attribute a certain set of characteristics to something called "human nature", then this set has to be attributable to all, or at least to most representatives of our species. But most people don't kill, most people don't steal, most people don't enslave. And I don't think that there is any evidence that killing, stealing and enslaving has been common before the first venerated rock.
Your metaphor with the arsenic seems to be quite tricky: Morality is attributable to humans and aspects of human life only. Religion is still a substantial aspect of human life, but arsenic is not. That is why arsenic will always be morally neutral, but we can't say the same for religion.
And of course, we can judge for the moral charge of religion only by its implication in our world. Do we have a choice? We can't unplug it even if we want. If religion stops interacting with our world, it won't be religion any more – it would be something else (or nothing at all).
As for Stephen Law, his main point is that religion is not always so useful. So when he says that its catalytic power is neutral, he is just being polite, nothing more.
I like the nuclear power metaphor to the point it shows how the world is trying to dispose of something only because it is potentially dangerous – no matter how useful this something is. And religion can only dream of being as useful as nuclear power.