Fraud and abuse are ubiquitous in modern digital ecosystems, manifesting across e-commerce,
social media, cloud computing, and telecommunications. While fraud typically centers on
financial deception and theft, “abuse” represents a broader, context-dependent set of challenges
that vary significantly by field—ranging from the exploitation of cloud infrastructure and
advertising click-fraud to behavioral toxicity on social platforms and the systemic manipulation of
search or ranking algorithms. Historically, the research and industrial communities have
addressed these challenges through a piecemeal approach, treating “credit card fraud,” “social
media misinformation,” and “account takeovers” as isolated domain problems. However, in today’s era, generative AI toolchains span across multiple domains in an unprecedented
manner, enabling automated, cross-platform attacks that blur traditional boundaries. Threat
actors now leverage common generative frameworks to create synthetic identities, realistic
phishing campaigns, and coordinated botnets that impact diverse sectors simultaneously. This
workshop aims to take a holistic lens on the detection and prevention of such fraud and abuse,
moving beyond domain-specific silos to identify universal patterns and scalable AI-driven
defenses.
The workshop is designed for a cross-disciplinary audience of researchers and practitioners:
● Computer Science Researchers: Experts in anomaly detection, adversarial machine
learning, graph neural networks (GNNs), and Large Language Model (LLM) security.
● Industry Trust & Safety Professionals: Teams from big tech, retail, gaming, and fintech
who manage platform integrity and user safety.
● Data Scientists and ML Engineers: Those building real-time detection systems that must
balance high-precision filtering with low-latency requirements.
● Ethics and Policy Researchers: Individuals studying the societal impact of AI-driven
abuse and the regulatory frameworks (e.g., AI Act, KYC) governing automated defense.