MLOps.community Fraud Detection in the AI Era // Rafael Sandroni // #301
13 snips
Apr 1, 2025 Rafael Sandroni, Founder and CEO of GardionAI, dives into the realm of AI security and fraud detection. He discusses the importance of establishing robust security guardrails to combat vulnerabilities like prompt injection attacks. The conversation highlights the zero trust framework as essential for safeguarding AI systems, particularly in the financial sector. Sandroni also contrasts traditional fraud detection with modern AI methods, shedding light on the evolving landscape of cybersecurity and the need for proactive measures.
AI Snips
Chapters
Transcript
Episode notes
AI Assistant Best Practices
- Thoroughly test AI systems, especially integrations, using a mix of AI and human testers.
- Implement guardrails to control outputs, prevent regressions, and ensure quality, especially in regulated markets.
Siri's Potential
- Demetrios Brinkmann expresses his preference for Siri as an ideal AI assistant abstraction layer.
- He wishes Siri could integrate deeper into apps, questioning the need for in-app AI assistants.
Nubank Experience
- Rafael Sandroni worked at Nubank, contributing to their transparent engineering blog.
- He built the basis of Nubank's AI assistant and later worked on real-time machine learning for fraud detection.