The algorithmic covenant: why AI-driven consent must not replace clinician responsibility.

Abstract

Health systems are increasingly deploying artificial intelligence (AI) tools to deliver procedural information, generate documentation, and support patient-facing "consent" workflows. Although such tools may improve aspects of an already deficient consent process, they create a risk of substitution: treating algorithm-mediated disclosure as ethically equivalent to clinician-led dialogue. This paper performs a conceptual analysis of the "algorithmic covenant" - the implicit expectation that automation will preserve the moral commitments of care. It draws on legal standards of material risk, sociotechnical scholarship on "moral crumple zones", and the tension between managerial and professional ethics. Current AI deployments risk facilitating "responsibility laundering." By prioritising system-level efficiency, institutions may use AI to standardise disclosure while leaving individual clinicians to absorb blame and legal risk for the resulting compliance artefact. This creates a misaligned responsibility structure where the institution controls the content, but the clinician bears the risk, potentially degrading consent into a digitised tick-box exercise. To preserve the ethical integrity of informed consent, AI must be governed strictly as an assistive technology. Policy and workflow design must explicitly assign responsibility for verifying patient understanding to named clinicians, ensuring that automation supports - rather than displaces - the clinician's non-delegable ethical responsibility for informed consent.