This session explores the transition from reactive moderation to proactive "Agentic Tooling," a model where autonomous AI agents leverage machine-readable policies (MRP) to perform deep, system-level analyses of platform health. We will dive into the architectural necessity of "co-design," where policy experts and engineers collaborate to build Domain-Specific Languages (DSLs) that translate complex legal nuance into deterministic, executable code. Additionally, we will examine how moving away from rigid, rule-based systems toward deep cross-functional collaboration (across policy, operations, and engineering) can achieve parity between human and automated violation detection. By modeling agentic systems after human reasoning steps, platforms can achieve a 10x to 100x reduction in cost-per-review while ensuring alignment between policy intent and enforcement action. Drawing on our collective experience in the safety and privacy domains, we will discuss how to successfully apply and scale these models.

Scam operations route across platforms, banks, telcos, and payment providers because the gaps between those sectors are where enforcement stops. That is a deliberate operating choice, and AI has made it cheap to run at scale. This session is about what happens when intelligence crosses those lines anyway. Joanne Hurley from Meta walks through a specific investigation end to end, including how Meta is using AI to increase the scale and impact of enforcement. Outtake, which submits signals into Meta's Fraud Intelligence Reciprocal Exchange (FIRE) on behalf of partners across sectors, picks it up from a vantage point outside any single industry. We close on what reciprocal sharing already delivers through FIRE and the Global Signal Exchange, how early in the attack chain it has to happen to matter, and what it would take to make this the norm rather than the exception.

Fraud is never just external; for modern marketplaces, risk exists across the entire ecosystem—in internal tools, outsourced operations, and the platform’s edges where external actors continuously test controls. Vendors and partners may also be exposed due to commercial pressure, turnover, or incentive misalignment, creating additional vulnerability points. This panel examines fraud as an end-to-end ecosystem challenge rather than a point solution or isolated Trust & Safety issue, exploring how it surfaces across three layers: in-house employees with privileged access, outsourced vendors and BPOs handling critical workflows, and external bad actors ranging from organized fraud rings to opportunistic users and toolkits probing for weaknesses.

Generative Artificial Intelligence (GenAI) is revolutionizing financial risk management by automating complex tasks such as report generation, synthetic data creation for testing, and cross-jurisdictional analysis in global institutions. However, its inherent challenges—including hallucinations, model opacity, and handling of sensitive data—introduce significant data security and privacy risks that strain existing governance frameworks. This session examines GenAI applications across key risk domains (market, credit, operational, liquidity, and model risk) using a case study of a hypothetical systemically important bank. It highlights privacy implications, such as potential data leakage in training or outputs, and alignment with global regulations including the EU AI Act and emerging privacy guidelines.

Artificial Intelligence has been in talk since data was considered as an asset. Using this data, prolific information has been extracted to gather detailed information and create datasets. Artificial intelligence is shaping the narrative of the world in the 21st century. It has become a driving force behind innovation, influencing various industries and redefining traditional practices. Artificial intelligence has revolutionized customer engagement strategies across industries, particularly Banking. This study explores the role of AI-based promotions platforms enhancing customer engagement, personalized promotions, and driving business growth. By leveraging advanced machine-learning algorithms and data analytics, banks can segment their customer base, analyze spending patterns, and design customized promotions that resonate individual preferences. This approach not only increases customer loyalty but also drives revenue growth and strengthens brand positioning. This customer-centric approach not only fosters greater customer loyalty but also contributes to increased revenue streams and strengthens brand positioning in a competitive market. This research aims to provide actionable insights for banking institutions to design and implement effective AI-powered promotional strategies while addressing ethical and operational considerations. It also examines the challenges faced by the banking sector in integrating AI, emphasizing the importance of transparency, accountability, and fairness in AI systems. The study explores how AI can enhance customer engagement, optimize marketing campaigns, and drive personalized financial services. Additionally, it highlights the need for robust regulatory frameworks and ethical guidelines to ensure that AI-driven initiatives align with both customer expectations and societal values. By addressing these elements, the research seeks to bridge the gap between technological advancements and responsible implementation in the banking sector.