数字信贷中负责任与可信人工智能原则

March 2026 Foreword Rajnil MalikPartner, AI GTM Leader PwC India AI has moved rapidly from being a conceptapproach, one which balances businessto a core driver of business operations andobjectives, technological advancementsinnovation. In digital lending, AI-poweredwith regulatory expectations and most decisions, enhancing customer experience,PwC is glad to collaborate with Dvaraand extending financial access to theResearch Foundation and the FinTechunderserved. However, as AI becomesAssociation for Consumer Empowermentdeeply embedded in lending operations,for developing this paper which presentsit becomes essential to minimise thea comprehensive view of how AI is can impact financial outcomes, principlessuch as fairness, accountability, and trustmust be built into AI systems from the verybeginning. This urgency is reflected in the As the digital lending landscape continueswhich reinforces many of the themesto evolve, we believe that organisations thatexplored in this paper, and highlightslead with RAI will be better positioned to AI in digital lending needs a multidisciplinary Foreword Indradeep GhoshExecutive Director AI is increasingly being integrated in theThe natural next step for each sectordesign and delivery of financial services.is to contextualise these sutras andA survey conducted by RBI’s Departmentoperationalise them through actionableof Supervision found that about a fifthpractices and this whitepaper is a step(20.8%) of the supervised lenders reportin that direction. The paper also containsusing AI in their operations and an evena web-based checklist of best practiceslarger proportion expresses an interestthat allows digital lenders to align theirin incorporating AI.1Stakeholders in theuse of AI in conformity with the mandate importance of implementing responsible andtool. It scores the strength of the existingtrustworthy AI practices as they understandAI safeguards of digital lenders, allowing The Indian policy discourse, first through theRBI’s FREE-AI Committee and then throughthe India AI Governance Guidelines, haspresented a rich framework of ‘RBI’s sevensutras’ to ensure that AI can be used in aresponsible and trustworthy manner. The We hope that you find the paper insightful. Foreword Sugandh SaxenaChief Executive Officer FinTech Association for Consumer Empowerment (FACE) From customer screening to credit-risklifecycle, lenders now have a structuredassessment, and from identifying fraud toway to locate responsibility among complexpricing, servicing and collection, automatedtech-stacks and distributed teams. Thedecisions bolster many elements inpaper supplements this identification withthe digital lending lifecycle. Since AI isconcrete practices to translate responsibleregularly retrained to work with third-partyand trustworthy AI (RTAI) from start (solutionvendors and human oversight, no singledesign, and model development) to finish The RBI regularly attempts to developframeworks to mitigate these issues, FACE regularly engages with its membershipincluding the recent FREE-AI Committeebase of 300+ FinTechs, and understandsReport which presents the guiding principlesthat FinTechs cannot have uniform capabilityof the seven sutras and recommendationsor intent. Thus, the paper’s distance As AI is increasingly being used in digitalconvert AI principles into operationallending to scale its operations and enhancestandards, norms, and checklists that thecustomer satisfaction, this whitepaper and been discussed in this white paper, thoughRTAI’s principles. We hope that you find thisits innate value lies in drawing attentionpaper to be useful and insightful. to upstream decisions. Since the paperexamines AI systems across the model’s Contents 06 Introduction 01Principles of RTAI08 02How RTAI can help digital lending 03Implementing RTAI in digital lending20 04Way forward21 Introduction The adoption of artificial intelligence isthem to improve account usage, promotesteadily increasing across various industries.budgeting, and deepen financial literacyThe financial sector, in particular, is atthrough relevant, customised and timelythe forefront of this movement, and iscontent. In the case of credit, fuelled byharnessing AI to realise considerablebig data, algorithms could do a bettercommercial value while simultaneouslyjob in predicting creditworthiness of thin- 1. Data processing abilities:Enhanceddata processing abilities, including theability to process qualitative and audio-based data could lead to a deeper needs and offering customised, relevantJust like the benefits, the risks could alsoand timely support in a customer-friendlyincrease with the model, affecting veryformat along with the customer journey.large number of customers at once. TheAI systems can also help financial servicedifficulties in explaining AI processes andproviders (FSPs) build stronger defensesimplementing complex algorithms could 2. Flexibility and scalability:AIsystems exhibit a high de