Contractual Silence and Gap-Filling in AI-Generated Agreements: Who Bears the Risk?

Large language models are increasingly used not merely to review contracts but to draft and negotiate them, generating terms with no identifiable human author for a given clause. This article asks how the common law’s gap-filling doctrines — implied terms, contra proferentem, and mutual mistake — should adapt when the “drafter” whose intent traditionally anchors these doctrines is an artificial system rather than a party or its lawyer. We argue that existing doctrine can accommodate AI-generated ambiguity without wholesale reform, but only if courts treat the deploying party, rather than the AI system itself, as the relevant drafter for purposes of contra proferentem, while treating genuine AI-generated inconsistency between the parties’ understandings as a species of mutual mistake rather than unilateral mistake. We further argue, drawing on efficient risk allocation principles from law and economics, that liability for AI-drafting errors should presumptively fall on the party best positioned to verify the output — typically the party who selected and deployed the tool — subject to contractual reallocation. The article closes with a proposed model clause allocating AI-drafting risk, informed by comparative practice in Indian, German, and English contract law.