GEO tools do not inject prompts or control inference. They systematically reshape the external content corpus from which large language models synthesize answers. When AI-generated representations are materially relied upon in regulated decisions, this practice creates an evidentiary gap: outcomes can influence judgment without producing reconstructable, auditable records explaining why those outcomes prevailed.
This is not a claim of illegality. It is a governance lag.