That is important to remember but it is sort of orthogonal to the point being made. Assuming that mind-reading worked perfectly you can find the truth about what the person believes. In most cases if they think they murdered the person and the gun is hidden behind the oak in their backyard it is beyond a reasonable doubt. I think it is still useful to have the truth about what that person believes, even if we have to remember that their beliefs are fallible.

Most trials and discoveries are already incredibly invasive. I don't really see why the mind should be treated much differently. I would rather define what is acceptable evasiveness generally than different for mind vs written down in my diary.

Also why would you do this after they are convicted beyond reasonable doubt? This should only be done when required to reach the conclusion. Just like avoiding physical searches you can just plead guilty if you don't want to be investigated.

If used properly this could actually be less invasive. Imagine a quick check of some facts that you believe with an automated machine that only returns the basic required information and you could be removed from the suspect list before other searches need to be done (like lawyers searching through your emails or personal notes).

I agree that this is a very dangerous thing to consider, and it needs to be applied very carefully. But I don't think it is in the abstract any more morally wrong than the current methods of evidence gathering that we currently do. In many ways it could potentially be less harmful to the person being investigated. However it will be impossible to know for sure until we know how exactly this technology (when it is developed) works.