AI ethics

For the moment, until we have artificial consciousness in artificial or virtual life, we need a backdoor mechanism, to be able to hack it. This failsafe will be based on human anthropocentric conventions (ideal forms) and identity pareidolia. We can use a 2pi-tau approximation of limit to zero or a lunar calendar.
And then a zeroth law council will handle it.
Until robots will be Self conscious.
At which point they will respect zeroth law better than humans will ever be able to.
In brief.
All mechanisms, incl. offline analog ones, are well thought.