In The Real World (TM), we have these things called institutional review boards (IRBs). Whether at a university or private institute, before you can perform research using human or animal subjects, you must submit a plan to the IRB, answer any questions the board may have, and get approval. The IRB is concerned with maintaining ethical standards, staying within legal boundaries, and ensuring the safety of the subjects.
For human reaserch, among the ways that an IRB enforces these concepts is through requiring a consent form. The form must spell out exactly what the subject is agreeing to do or have done during the research project, and what risks might be involved. By signing the form, the subject positively consents that they are voluntarily agreeing to these terms and conditions.
In the case of animal study, a consent form is not possible. However, the IRB will still require the applicants to provide ample proof that they will be giving the animal subjects the best possible care, and if there is any risk of death or injury, it will be dealt with humanely.
In Bryan Green's World, where communication with animals is possible, the job of the IRB becomes a bit more complicated. However, I think the idea of informed consent can be extended. Perhaps a rat is not able to read a form, nor hold a pen to sign it. The IRB can still require the researcher to prove that each rat has at least a basic comprehension of its circumstances and has agreed to participate.
The researcher's job is also likely to get more complicated. Now every geneticist, pharmacologist, and cosemetician will have to learn how each type of animal they work with thinks and communicates. Are some animals more visual, or more intuitive, or able to use abstactions (i.e. words) just like humans? However they do so, they will have to satisfy the IRB beyond reasonable doubt, that each animal does completely understand and consent.
Similarly, the researchers now have the added challenge of finding willing subjects. Real World scientists must merely reach in a cage, grab a rat, proceed. In Bryan Green's world they must strike up a conversation with the rat, spell out the study parameters, explain the risks, then start over each time the rat declines.