The Internet Trust Problem

Most fraud teams set alert thresholds with a finger in the wind. Kunal Datta thinks there’s a better way.Is it machine learning or rules? Kunal Datta, Chief Product Officer at Unit21, says you don’t have to choose. Fresh from Money20/20, Kunal joins Matt Brady to talk about why “signals over scores” beats a mystery fraud number, how AI can tune rules from as few as 10 examples, and why trust in AI agents has to be earned step by step. We also cover how scammers use AI to run a thousand pig butchering scams at once, and why using AI in fin crime is a moral duty.Chapters:00:00 Setting alert thresholds with a finger in the wind00:59 Welcome and Money20/20 recap02:19 Where machine learning earns its keep04:39 Where machine learning starts to crack07:46 Rules: clear logic and catching new behaviour08:45 Getting the best of rules and AI12:45 Signals over scores15:09 Gut feel, decision instructions and real vs written SOPs17:32 Building trust in AI: progressive autonomy21:54 Who’s moving fastest: crypto vs banks24:50 The risk of leaner teams and AI doing the legwork30:10 How scammers are using AI34:07 Quick fire: ML, rules and gen AI34:42 Agent payments and new types of disputes36:27 The Anthropic IPO38:06 Where to find KunalResources:Kunal’s article, “A finger in the wind: How institutions decide which crimes to ignore”: https://www.unit21.ai/blog/a-finger-in-the-wind-how-institutions-decide-which-crimes-to-ignoreUnit21: https://unit21.aiKunal on LinkedIn: https://www.linkedin.com/in/kunal-datta/Watch and listen:YouTube: https://www.youtube.com/channel/UC8Gm5R039pPnJqS27QdvYlw/Apple Podcasts: https://podcasts.apple.com/us/podcast/leading-detection/id1854719167More from Sedulo Search:Website: https://sedulo.ioLinkedIn: https://www.linkedin.com/company/sedulo-search/Follow Matt Brady: https://www.linkedin.com/in/mattbrady-sedulo/Sedulo Search is a retained recruitment firm that places product, ML, engineering and solutions talent in fraud prevention. If you’re hiring or looking for your next role in fraud, get in touch: https://sedulo.io

Over half of job applicants could be fake, and bots now make up more than 70% of internet traffic. Persona co-founder and CTO Charles Yeh joins Matt to talk about deepfakes, digital twins and how we prove who's real online. Plus, why the future of fraud prevention is less about catching bad actors and more about protecting the humans.► Subscribe to the Leading Detection Podcast: @sedulo_search ► Listen on Spotify: https://open.spotify.com/show/6ci2UpvNNmaLBrXimQ9pe3?si=pF2pE-0GS-6woZCZuhnejQ► Listen on Apple Podcasts: https://podcasts.apple.com/us/podcast/leading-detection/id1854719167More from Sedulo Search - Technical Talent For Fraud Prevention:► Website: https://www.sedulo.io► LinkedIn: https://www.linkedin.com/company/sedulo-search/► Follow Matt Brady: https://www.linkedin.com/in/mattbrady-sedulo/Sedulo Search is a retained recruitment firm that places product, ML, engineering, and solutions talent in fraud prevention. If you're hiring or looking for your next role in fraud, get in touch: https://www.sedulo.io/fraud-insights#collaboration SummaryIn this episode, Charles, CTO at Persona, discusses the evolving landscape of digital identity verification, fraud detection, and the challenges posed by increasingly sophisticated AI-generated fake identities and videos. Learn about the latest techniques in combating candidate fraud, deepfakes, and the future of digital IDs.Key TopicsThe rise of candidate fraud and synthetic identitiesMultimodal fraud detection techniques combining computer vision and social profile analysisThe complexity of deepfake and video manipulation attacksLayered fraud attack vectors and adaptive defensesThe future of digital IDs and the shift towards digital identity verificationBalancing security, user experience, and trust in online platformsThe impact of AI-generated content on fraud detectionStrategies for reducing friction for legitimate users while blocking fraudstersThe importance of ensemble models and multimodal signals in fraud detectionThe societal and technical challenges of moving towards fully digital IDsChapters00:00 Introduction to digital identity challenges00:22 The complexity of candidate fraud01:24 Types of identity fraud and synthetic identities02:33 Multimodal fraud detection techniques03:12 Balancing security and user experience04:07 Risks of candidate fraud for companies05:10 Attack vectors: identity muling and digital twinning06:49 The rise of sophisticated deepfakes11:49 Using ensemble models for fraud detection15:20 Layered attack techniques and countermeasures20:01 Segmentation and risk-based verification22:13 Reducing friction for legitimate users27:04 The future of digital IDs and AI advancementsKeywordsdigital identity, fraud detection, deepfakes, computer vision, AI, digital IDs, identity fraud, biometric verification, cybersecurity, online safety#DigitalIdentity #BiometricVerification #Deepfakes #FraudDetection #Cybersecurity

After two decades fighting fraud at Google, Facebook, and Square, Kevin Lee (now CTO, Trust & Safety at Sift) sits down to explain why he builds every system with one customer in mind: the fraudster. He talks about silos, the coming headcount reckoning, and why trust & safety is quietly becoming a company's biggest growth lever.► Subscribe to the Leading Detection Podcast: @sedulo_search ► Listen on Spotify: https://open.spotify.com/show/6ci2UpvNNmaLBrXimQ9pe3?si=pF2pE-0GS-6woZCZuhnejQ► Listen on Apple Podcasts: https://podcasts.apple.com/us/podcast/leading-detection/id1854719167More from Sedulo Search - Technical Talent For Fraud Prevention:► Website: https://www.sedulo.io► LinkedIn: https://www.linkedin.com/company/sedulo-search/► Follow Matt Brady: https://www.linkedin.com/in/mattbrady-sedulo/Sedulo Search is a retained recruitment firm that places product, ML, engineering, and solutions talent in fraud prevention. If you're hiring or looking for your next role in fraud, get in touch: https://www.sedulo.io/fraud-insights#collaboration Kevin Lee shares his extensive experience in trust and safety, exploring how fraud and abuse have evolved in the digital age, the importance of holistic approaches, and the future of AI in combating digital threats.Key TopicsEvolution of trust and safety over two decadesExpansion of fraud vectors beyond payment fraudThe role of policy and regulation in trust and safetyAdversarial mindset of bad actors and system designThe importance of holistic, integrated fraud prevention systemsThe impact of AI and data fragmentation on trust and safetyTalent and organisational evolution in trust and safety teamsThe role of vendors and industry consolidationFuture predictions for trust and safety in a rapidly evolving landscapeChapters00:00 Introduction and Kevin's background01:16 The evolution of fraud from credit card theft to account takeover02:07 The interconnected nature of abuse vectors03:16 Growth and expansion of trust and safety concerns04:20 Policy, regulation, and industry accountability05:34 The convergence of fraud and trust safety07:15 Adversarial mindset and system design for fraud prevention08:34 Organizational silos and the need for integration10:00 Industry collaboration and sharing best practices11:46 Accountability and blame in trust and safety13:50 Human expertise and AI in fraud detection16:54 Limitations of technology and the human element18:51 Differentiating real from fake in AI-generated content20:55 Data fragmentation and its impact on trust and safety22:18 Talent evolution and team skills in trust and safety24:35 From reactive to proactive fraud prevention27:32 The role of technology versus people in trust and safety28:45 Building a trust and safety stack for the future34:00 Industry consolidation and vendor ecosystem35:27 Near-term and long-term predictions for trust and safety#TrustAndSafety #AI #DigitalTrust #FraudPrevention #AccountTakeover

Shuman Ghosemajumder, ex-Google and former CTO at Shape Security, now founder of REKEN, joins Matt to explain why the internet was never built for trust, and how AI is supercharging both scams and the tools to fight them. From deepfake phishing to "agents talking to agents," they explore what a trustworthy internet could look like and why on-device AI may be the answer.► Subscribe to the Leading Detection Podcast: @sedulo_search ► Listen on Spotify: https://open.spotify.com/show/6ci2UpvNNmaLBrXimQ9pe3?si=pF2pE-0GS-6woZCZuhnejQ► Listen on Apple Podcasts: https://podcasts.apple.com/us/podcast/leading-detection/id1854719167More from Sedulo Search - Technical Talent For Fraud Prevention:► Website: https://www.sedulo.io► LinkedIn: https://www.linkedin.com/company/sedulo-search/► Follow Matt Brady: https://www.linkedin.com/in/mattbrady-sedulo/Sedulo Search is a retained recruitment firm that places product, ML, engineering, and solutions talent in fraud prevention. If you're hiring or looking for your next role in fraud, get in touch: https://www.sedulo.io/fraud-insights#collaboration In this episode, Shuman discusses the evolving trust infrastructure of the internet, the impact of AI on online trust, and how to build a safer, more trustworthy digital environment. We explore the challenges of AI-generated content, fraud prevention, and future solutions for a zero-trust internet.Key Topics• The impact of AI on online trust and misinformation• The evolution of internet trust from military to public use• Fraud and cybercrime targeting large user bases• The importance of trust signals and indicators• AI-generated content and its implications for trust• Future solutions for a zero-trust internet• On-device analysis for secure communication• Building trust through transparency and privacyChapters00:00 Introduction and guest background00:41 Why trust on the internet is a growing concern01:00 AI's role in exacerbating trust issues02:20 The lack of trust infrastructure on the internet03:19 Historical trust in the early internet07:41 Different types of distrust and trust evolution09:04 The role of encryption in building trust10:30 Visual cues and user awareness of fraud12:13 Banner blindness and user behaviour14:07 The ineffectiveness of traditional phishing training17:23 The need for AI-driven trust solutions18:46 Envisioning a trusted internet with AI and trust indicators22:10 The importance of trust indicators for AI content24:41 Media and AI-generated content30:55 Upgrading communication infrastructure for trust34:21 On-device analysis and real-time protection36:06 Building a safer internet for humans37:40 The concept of a zero-trust internet38:46 The mindset for a trust-based digital future40:43 Conclusion and future outlook#Cybersecurity #AISafety #AIGeneratedContent #OnlineFraud #DigitalSecurity

Michael Jabbara, SVP at Visa, joins the show to break down VAMP (Visa's Acquirer Monitoring Program): what it is, why Visa built it, and what merchants and acquirers actually need to do to stay compliant. He shares real data showing that fixing fraud and dispute issues doesn't hurt growth; it actually boosts approval rates and helps businesses grow faster.► Subscribe to the Leading Detection Podcast: @sedulo_search ► Listen on Spotify: https://open.spotify.com/show/6ci2UpvNNmaLBrXimQ9pe3?si=pF2pE-0GS-6woZCZuhnejQ► Listen on Apple Podcasts: https://podcasts.apple.com/us/podcast/leading-detection/id1854719167More from Sedulo Search - Technical Talent For Fraud Prevention:► Website: https://www.sedulo.io► LinkedIn: https://www.linkedin.com/company/sedulo-search/► Follow Matt Brady: https://www.linkedin.com/in/mattbrady-sedulo/Sedulo Search is a retained recruitment firm that places product, ML, engineering, and solutions talent in fraud prevention. If you're hiring or looking for your next role in fraud, get in touch: https://www.sedulo.io/fraud-insights#collaboration SummaryIn this episode, we explore the Visa Acquire Monitoring Program (VAMP), a crucial initiative to combat outlier fraud and dispute behaviour in digital payments. Michael Jabbara from Visa shares insights on how VAMP works, its impact on growth and security, and how merchants and acquirers can leverage it for long-term success.Key TopicsVAMP's purpose and originHow VAMP detects outlier fraud and disputesImpact of VAMP on merchant growth and securityThresholds and remediation strategies in VAMPEvolving threats in digital payments and future predictionsChapters00:00 Introduction to the episode and guest02:22 What is VAMP and its origins03:23 Design and evolution of the Visa Acquire Monitoring Program04:18 Thresholds and incentives for upstream controls05:21 Why digital payments growth necessitated VAMP06:11 Early results and learnings from VAMP implementation07:29 Impact of thresholds on merchant growth and risk08:43 Remediation strategies and long-term benefits10:19 False positives and improving approval rates11:10 The relationship between risk, security, and growth12:44 Operational differences before and after VAMP13:31 Explaining VAMP to merchants and acquirers15:06 Control frameworks and best practices for fraud prevention16:52 Global consistency and regional adjustments18:36 The importance of evolving fraud prevention programs20:25 Cardholder impact and fraud behavior22:27 Future trends and program evolution24:39 Advising on upstream controls and best practices28:48 Summary of VAMP's benefits for merchants and acquirers29:56 Final thoughts and resources for further information#VisaVAMP #FraudPrevention #DigitalPayments #PaymentSecurity #RiskManagement

Hanjo Seibert, Managing Director of Risk & Compliance and TED speaker, joins Leading Detection to break down why fraudsters are innovating faster than the institutions meant to stop them — and why fixing it comes down to data, agile teams, and bold leadership, not just technology.► Subscribe to the Leading Detection Podcast: @sedulo_search ► Listen on Spotify: https://open.spotify.com/show/6ci2UpvNNmaLBrXimQ9pe3?si=pF2pE-0GS-6woZCZuhnejQ► Listen on Apple Podcasts: https://podcasts.apple.com/us/podcast/leading-detection/id1854719167More from Sedulo Search - Technical Talent For Fraud Prevention:► Website: https://www.sedulo.io► LinkedIn: https://www.linkedin.com/company/sedulo-search/► Follow Matt Brady: https://www.linkedin.com/in/mattbrady-sedulo/Sedulo Search is a retained recruitment firm that places product, ML, engineering, and solutions talent in fraud prevention. If you're hiring or looking for your next role in fraud, get in touch: https://www.sedulo.io/fraud-insights#collaboration Episode summaryIn this episode, Hanjo Seibert discusses the evolving landscape of fraud detection and prevention, the impact of technological innovation, and strategic approaches for financial institutions to stay ahead in the fight against fraud.Key TopicsEvolution of fraud detection over the past decadeImpact of generative AI and automation on fraud schemesStrategies for agile implementation of fraud prevention toolsThe importance of organisational culture and leadership in innovationDifferentiating between fraud and AML challengesThe significance of data-driven insights in fraud detectionClient interaction design as a fraud prevention measurePredictions for fraud trends in the next 12-24 monthsChapters00:00 Introduction and guest background02:25 Evolution of fraud detection over 10 years04:37 Challenges of outdated solutions and implementation timelines06:02 Shortening timelines as an advantage for market winners07:27 Fraud innovation versus regulatory constraints08:47 Fraud as a sandbox for broader compliance innovation10:17 External factors and social engineering in fraud11:38 Designing customer journeys with fraud controls13:53 Real-time fraud detection challenges17:29 Industry questions on scams and client education19:33 Leveraging technology and AI in fraud schemes22:49 Advising clients on fraud and technology strategies25:10 Growing fraud problem and industry awareness27:03 Barriers to innovation: change management and culture28:25 Designing future-proof fraud prevention architectures30:28 Outlook for the next 12-24 months32:21 Leadership and organisational change for fraud innovation34:48 The role of executive leadership in driving change36:00 Predictions for fraud trends and technological advances38:03 Closing remarks and contact information#FraudDetection #FinancialTechnology #AI #AML #FraudPrevention

Deepanker joins us to discuss the evolution of identity verification, the challenges of fraud in digital spaces, and the future of adaptive security measures. This episode explores how technology and behavioural signals are reshaping trust and security in online transactions. Key TopicsThe evolution of identity verification from face-to-face to digitalThe impact of technology on fraud methods and detectionThe importance of behavioural signals in securityThe role of context and continuous verification in fraud preventionEmerging risks like age assurance and workforce fraudChapters00:00 Introduction to Deepanker Saxena and IDV02:04 The impact of technology on fraud methods04:22 The rise of remote fraud and technological countermeasures06:30 The importance of multi-vector verification09:09 The role of context in fraud detection11:21 The significance of platform-wide context and history13:42 Predictive models and understanding user behaviour15:20 Continuous identity verification throughout the customer lifecycle17:12 Seamless user experience with subtle signals20:10 The rise of age assurance and its importance21:26 Risks associated with digital age verification22:56 Emerging risks in workforce verification24:38 Fraud rings and sophisticated workforce fraud tactics27:04 The ongoing evolution of fraud tactics27:59 The concept of agentic fraud and prompt injections29:31 The importance of binding identity to agents32:20 The future of identity and agent-based verification33:19 Building a comprehensive ecosystem of identities34:14 The impact of agent-based systems on daily life35:20 Emerging trends and future challenges in identity verification#cybersecurity #techpodcast #digitalidentity #fraudprevention #biometrics

In this episode, Colin Parsons from Nasdaq Verafin discusses the current state of financial crime, the impact of AI on fraud, and innovative strategies for detection and prevention. Discover how industry leaders are adapting to rapid changes and leveraging new technologies to combat fraud effectively.Key TopicsRise in global financial crime to 4.4 trillion annuallyIncrease in authorised scam losses by 19.3% over two yearsShift from human-scale to machine-scale fraud driven by AIImportance of network data and consortium insights in detectionProactive fraud prevention beyond real-time transaction analysisRole of agentic AI in enhancing investigator efficiencyThe impact of AI models and agentic AI workforce in fraud detectionFuture trends: more scams, credential stuffing, and vulnerabilitiesChapters00:00 Introduction and guest introduction00:44 Current mood among fraud teams and industry uncertainty02:07 Overview of the global financial crime report03:06 Rise in financial crime and fraud losses04:40 Impact of AI on fraud and scams06:11 Shift to machine-scale fraud and deep fake scams07:09 Factors driving increased sophistication and volume of fraud09:07 Shift from cyber to human scam focus10:32 Education and friction in fraud prevention12:22 Using data and context to identify scams earlier15:08 Network approach and consortium data in fraud detection18:22 Real-time vs proactive fraud prevention strategies19:13 Leveraging AI to filter signals and reduce noise20:40 Unified view for investigators and AI models21:20 Role of agentic AI in fraud investigation23:36 Adoption and surprises with agentic AI workforce25:38 Combining AI with network insights for effectiveness26:25 Predictions for the next 6-12 months in fraud28:11 Resources and contact information for further learning#financialcrime #fraudprevention #ai #cybersecurity #fintech

In this episode, Eugeny Malyutin from SUMSUB discusses the realities and hype around AI and ML in fraud prevention, exploring how fraud tools have democratised, the current state of AI in fighting fraud, and best practices for deploying AI systems in regulated environments.Key TopicsThe democratisation of fraud tools and its implicationsCurrent state and future of AI and ML in fraud detectionBest practices for deploying AI in regulated environmentsThe role of human oversight in AI-driven fraud preventionChallenges and opportunities with generative AI and deepfakesChapters00:00 Introduction to the episode and guest03:06 Hype versus reality in AI-driven fraud agents08:00 Differences in applying ML in fraud versus other industries12:04 Building a modern, multi-model fraud detection system14:15 Speed of fraud evolution and detection challenges15:12 Data versus models: what is more important?16:22 Detecting new fraud patterns with data and AI17:02 Automating data labeling and improving models18:33 Balancing fraud detection and customer experience21:46 From demo to production in regulated environments22:17 Regulatory requirements for AI in fraud detection23:19 The importance of infrastructure and continuous evaluation24:11 What typically breaks when scaling AI systems25:41 Reducing confidence errors in large language models28:35 Generative AI's role in fraud detection and analysis29:50 Using AI for document analysis and environment configuration33:22 The hype around generative AI and its limitations34:19 Hiring for AI and ML in fraud prevention36:44 Onboarding new talent from other domains38:20 Eugeny's outlook for AI in the second half of the year#AI #Cybersecurity #FraudDetection #Deepfakes #GenerativeAI

In this episode, Pedro Sanzovo shares his journey into fraud prevention, the evolving role of AI, and how data-driven strategies are shaping the future of anti-fraud technology. Discover insights on balancing AI advancements with fraud defence and the importance of curiosity in this dynamic field.Key topicsPedro's journey into fraud preventionThe role of open banking in fraud detectionAI advancements and challenges in fraud preventionData assets and their impact on fraud defenceThe future of fraud products and industry trendsChapters00:00 Introduction and Pedro’s journey into fraud02:04 Starting in AML and fraud prevention at Palantir04:10 Founding a fraud prevention company in Brazil06:05 Acquisition by Signified and role in fraud at checkout07:46 Data infrastructure and the evolution of fraud detection09:58 Open banking’s influence on fraud prevention12:00 The importance of quality data and layered defences13:51 Balancing AI advancements with fraud prevention challenges15:48 Technologies like deepfakes and evolving fraud tactics18:00 The ongoing battle against fraud and AI’s role20:02 Leveraging AI to empower investigators and improve defences21:57 Transforming fraud decisioning with sequential models24:06 Change management and industry adoption of new tech26:06 The future of fraud as a growth enabler28:01 The blurring lines between fraud, credit, and marketing29:57 Emerging trends: first-party fraud and industry evolution32:05 Final thoughts and Pedro’s advice for aspiring fraud professionals#FraudPrevention #AIInFraud #DataDrivenSecurity #OpenBanking #FraudDetectionStrategies