THE IDENTITY THE ALGORITHM FABRICATED: BANKING FRAUD WITH SYNTHETIC PERSONS AND THE DUTY OF SECURITY

Abstract

Generative artificial intelligence has changed the cost and quality of forging identity signals: documents, faces, voices and behaviours can be produced with high coherence, and injection attacks insert forged media directly into the verification flow. This article examines to what extent the algorithmic fabrication of identities alters the content of the banking duty of security in the opening, validation and authentication of accounts and transactions, and which technical and evidentiary criteria distinguish a service defect, attributable to the institution as internal fortuity, from external fraud lacking a causal link with the banking infrastructure. The research is qualitative, exploratory and legal-dogmatic, combining normative analysis of the Consumer Defence Code, the General Personal Data Protection Law and the regulation issued by the National Monetary Council and the Central Bank of Brazil, an examination of precedents of the Superior Court of Justice, a narrative interdisciplinary review of the information security literature and of identification studies, and the use of foreign technical standards as persuasive reference, with research closed on 9 September 2026. Results indicate that synthetic fraud shifts the defect from documentary falsity to the architecture of trust, that liability remains strict within the limits of internal fortuity without being automatic, and that the legal order already contains the elements of a proportionate, dynamic, systemic and demonstrable duty of identity security. As its contribution, the article proposes the identity trust chain, the LACRE Matrix and the Minimum Authenticity Evidence Package.