Beyond Consent: Contestability and Recourse for Generative AI Identity Harm

Generative artificial intelligence can reproduce, transform, or simulate features that people use to identify themselves and earn a living, including voices, likenesses, writing styles, and other recognizable traits. Consent is an important safeguard, but it is not a complete ethical response once an identity related harm has already occurred. A person may never have been asked, may have been unable to give meaningful permission, or may face a harmful output after an apparently valid permission was obtained. This paper develops a rights based account of what affected people are owed after such harm. It argues that ethical governance must supply five linked capabilities: notice, intelligible explanation, contestation, timely remedy, and institutional learning. These capabilities are operational rather than merely aspirational. They require clear pathways to identify synthetic content, obtain a human review, preserve relevant evidence, correct or remove harmful material when justified, and learn from the incident. The paper distinguishes legal compliance from ethical adequacy: transparency duties and consent mechanisms can create useful information, but they do not by themselves allocate responsibility or guarantee recourse. The proposed RECOVER framework offers a practical way for providers, deployers, platforms, and organizations to design post harm governance around recognition, explanation, challenge, outcome, verification, escalation, and repair. Preprint; not peer reviewed.