We formalize strategic signal corruption (deception) as entropy injection into communication channels between cooperating agents. Using Shannon information theory and Landauer's principle, we prove three results. First, deception imposes a quantifiable decision cost on the receiver, proportional to the residual variance of the payoff-relevant parameter induced by the corrupted signal. Second, verification against deception requires redundant observations whose energy cost diverges as the corruption rate approaches the channel capacity limit. Third, pervasive signal corruption across a network of agents collapses cooperative surplus at a critical corruption rate, yielding a sharp network-collapse threshold. A key consequence is that no costless deception exists: for any non-zero corruption rate, at least one of the decision cost or the verification cost is strictly positive. Honest communication is the unique efficiency-maximizing regime. The results connect informational entropy injection to the physical friction framework developed in the first paper of this series, unifying overt boundary violations and covert signal corruption under a single cost accounting.