{"entity":{"id":"idea-data-ai-liability-safe-harbour","kind":"idea","name":"A liability framework for clinical AI: safe harbour for clinicians, liability for makers","aka":[],"tldr":"Make clear who is responsible when an AI tool contributes to a mistake: protect doctors who use approved tools as intended, and hold makers responsible for the tool's performance.","summary":"Liability uncertainty holds hospitals and clinicians back from adopting AI: the clinician may bear responsibility for a tool they cannot inspect. The proposal is legislation or regulatory guidance establishing a safe harbour for clinicians who follow a registered, validated model within its labelled use, coupled with product-liability accountability for developers for performance within the labelled use, and a no-fault compensation scheme for patients harmed by AI errors, as exists for vaccines in several countries.","asOf":"2026-09-08","links":[{"label":"Bottleneck evidence (AI that is built but not validated or deployed): Wu et al., How medical AI devices are evaluated: limitations and recommendations from an analysis of FDA approvals (Nature Medicine 2021)","url":"https://doi.org/10.1038/s41591-021-01312-x"}],"tags":[],"related":["idea-data-clinical-ai-model-registry"],"cancers":[],"sections":["ai-computation"],"technologies":[],"targets":[],"drugs":[],"companies":[],"institutions":[],"pathways":[],"terms":[],"trials":[],"people":[],"bottlenecks":["b-ai-validation"],"keyPapers":["paper-wu-nat-med"],"journals":[],"dependsOn":[],"notes":[],"hypothesis":"A clear liability framework will increase adoption of validated AI tools by clinicians and reduce defensive overriding, without an increase in patient harm, in jurisdictions that adopt it versus those that do not.","rationale":"Liability shields (Good Samaritan laws, vaccine injury compensation) have changed professional behaviour where uncertainty deterred beneficial action; surveys of clinicians rank liability among the top barriers to AI use.","test":"Compare AI adoption and override rates in a jurisdiction that adopts the framework with matched jurisdictions over three years; track claims and compensation cases.","maturity":"speculative","actor":"policy","cost":"small","horizonYears":3},"route":"/ideas/idea-data-ai-liability-safe-harbour/","neighbours":{"idea":[{"id":"idea-data-clinical-ai-model-registry","kind":"idea","name":"A public registry of every AI model used in cancer care","route":"/ideas/idea-data-clinical-ai-model-registry/"}],"section":[{"id":"ai-computation","kind":"section","name":"AI & Computation","route":"/fronts/ai-computation/"}],"bottleneck":[{"id":"b-ai-validation","kind":"bottleneck","name":"AI that is built but not validated or deployed","route":"/bottlenecks/b-ai-validation/"}],"paper":[{"id":"paper-wu-nat-med","kind":"paper","name":"How medical AI devices are evaluated: limitations and recommendations from an analysis of FDA approvals","route":"/key-papers/paper-wu-nat-med/"}]}}