This doesn't help you for those questionable entities. There is no sharp dividing line. Information theory can be very useful for detemining molecular (particularly DNA) meaning within specific contexts, but for a definition of life, it becomes inclusive of things like prions or even artificial life systems.
However, when we are speaking of Shannon applied to molecular biology we are not speaking of Shannon applied to artificial life. There is no reason to confuse the two applications - or either of them with other applications of Shannon, such as telecommunications and computer technology.
Also, I strongly disagree with the statement that the Shannon model does not provide a sharp dividing line. The theory is not at all concerned about the value or meaning of the message or whether the communication is taking place on a computer or in a molecular machine.
IOW, applied to the definition of life v. non-life/death the Shannon model gives a bright line of distinction between life and non-life/death. Where successful communications occur in nature, there is life. When there is no successful communications in nature there is death or non-life.
The theory is elegant, ideologically neutral and requires no subjective biochemical delineations as other definitions require. Because it is math and does not address the value or meaning of the message at all, Shannon-Weaver is not concerned whether prions are only proteins, whether mycoplasmas have no cell walls, whether viroids are RNA without a protein coat. They all fit within the communication model. The model addresses source, message, encoder, channel, decoder, receiver and noise. Noise can result in a miscommunication of an intended message and therefore, a malfunction (or perhaps improvement) in the molecular machine.
For Lurkers (taken mostly from the Schneider website):
A molecular machine is a single macromolecule or macromolecular complex which performs a specific function for a living system, is usually primed by an energy source, dissipates energy as it does something specific (pays the thermodynamic tab), `gains' information by selecting between two or more after states. Molecular machines are isothermal engines.
For more information on the subject:
Adami: Information Theory in Molecular Biology