白皮书：伦理人工智能的五大支柱：金融机构合规与负责任使用人工智能的操作指南。

A playbook for compliant andresponsible use of AI in financialservices organizations. Table of contents Page 01Governing Ethical Use of GenAI 03 The AI Center of Excellence04 Five Pillars of Ethical AI 04 051. Compliance and Regulatory Adherence 062. Policies and Standards 093. Risk Management 12 Be a Trailblazer in Ethical AI How Amdocs Can Help 12 Governing EthicalUse of GenAI Why Prioritize Ethical AI? Reduce exposure to risk:When AImodels go wrong, legal, operational,and reputational damage can result.Minimize this risk with robust policiesand procedures. As generative artificial intelligence (GenAI)increasingly pervades the technology and businesslandscape, governing ethical use across theenterprise is a critical consideration. Dedicatedregulatory measures are only just beginning toemerge so, currently, industry is largely responsiblefor defining responsible, compliant, and safe useof the technology. Wherever your business is onthat journey, this playbook provides a frameworkto ensure you are aware of, and able to deal with,potential pitfalls. It centers on five pillars ofethical AI: compliance and regulatory adherence,policies and standards, risk management, rolesand responsibilities, and transparency andexplainability. Demonstrate industry responsibility:If industry fails to set a precedentfor ethical use of AI, authorities aremore likely to take a heavy-handedapproach with future regulations. Get ahead of regulations:AI policiesand standards lay foundations forfuture compliance, saving time,money, and inconvenience when legalrequirements emerge. Nurture responsible use from theground up:Build enterprise-wideawareness and understanding,helping employees embrace ethical AInow to avoid problems later. AI has the power to deriveenormous business valuefrom data, but withoutcareful governance it cancause great harm. Be at the forefront of ethical AI:Earn competitive advantage as a leader in responsible use of AI,fostering trust with customers andother stakeholders. Why AI Needs Tailored Governance Many financial services enterprises consider AI thenext step in data analytics maturity. Consequently,there’s a common assumption that existing datagovernance can extend to AI. Yet this approachdoesn’t offer the necessary scale and precision. AIbrings new risks surrounding potential misuse ofdata as well as issues linked to the indeterministicnature of the large language models (LLMs) whichGenAI is based on. For instance, LLMs may generatedifferent responses to the same input due to theirprobabilistic sampling methods and the inherentrandomness of the decoding process. That’s whyAI governance requires a tailored approach and ahigh level of scrutiny, particularly in the financialservices sector. The AI Center of Excellence When adopting rapidly evolving technologies there can be a fine line between opportunity and risk,between confidence and loss of trust, and between success and failure. Amdocs has a long history helpingorganizations navigate new technology pathways and transformations. To maximize success in AI adoption,we advocate an AI Center of Excellence (CoE). The AI CoE establishes and oversees implementation of standards, tooling, best practices, policies, procedures,monitoring, education and more. Together these measures foster ethical use of the technology. In thisplaybook we detail five pillars of ethical AI which provide a framework for the CoE. We illustrate potentialpitfalls, and outline measures the CoE can take to mitigate risks. Our goal is to help you master responsibleuse of AI at scale. Five Pillars of Ethical AI Raise the bar on AI governance with an AI CoE that cultivates best practices across theenterprise. 1. Complianceand RegulatoryAdherence Understand Existing Regulations,Monitor Emerging Requirements The regulatory landscape for AI is less maturethan that for areas such as the use andstorage of personal data. Nevertheless,dedicated requirements are beginning toemerge. Monitoring regulatory discussions anddevelopments at a global and cross-industrylevel is vital so the CoE can keep one stepahead of future requirements. Potential Pitfall: Chatbots BreachingData Rules Regulatory obligations should be carefullyconsidered throughout the development ofGenAI-powered tools. For instance, chatbots’use of personal data must comply with rulessurrounding consent and disclosure. TheCoE can reduce the risk of non-complianceby identifying existing regulations thatrelate to GenAI use cases, as well asmonitoring emerging regulatory trends andrequirements. Organizations must also adhere to existingstandards and guidelines that are relevantto the implementation of AI systems. Dataprotection regulations are a case in point.Regulations from bodies such as the FinancialIndustry Regulatory Authority (FINRA) areapplicable to some aspects of AI use, as isthe EU’s general data protection regulation(GDPR). Emerging AI Regulations The AI Act Regulation (EU) 2024/16891, the