Back to Fraud Forward

Bonus Episode — Building Smarter, Not Harder: Using AI to Eliminate Fraud’s Busy Work with Ben Graf

December 12, 2025
Hailey Windham
HOST
Fraud Forward, Sardine
Benjamin Graf
Director of Trust & Safety
Listen on
YouTube
Apple Podcasts
Spotify
Pocket casts
Overcast
Share
Share this episode

In this bonus episode of Banking on Fraudology, powered by Safeguard, Hailey Windham talks with Ben Graf, a self-taught AI expert in the neobank space. Ben embodies the curiosity and courage driving the next wave of transformation in fraud fighting.

The conversation explores what it really looks like to learn AI from the ground up, emphasizing that the future of fraud prevention is not about replacing people, but about empowering them through technology.

Key takeaways: AI, innovation, and fraud-fighting empowerment

Using AI to learn AI

  • Ben explains how he used a variety of LLM tools, including ChatGPT, Claude, and Gemini, as coaches or mentors. He spent hours experimenting to understand their capabilities, consistency, and how to prompt them effectively. 
  • This approach helped him translate technical concepts and practices, such as data analysis, SQL, and JavaScript, into actionable knowledge for his team, breaking down communication barriers. 
  • The hardest part was knowing where to start, but the key realization was that “something is better than nothing,” and that compounding knowledge quickly lowers barriers.

Practical AI applications for eliminating busy work

  • Making teams more efficient: AI should be used to reduce manual effort and help professionals focus on higher-impact, strategic work.
  • Automating document verification: AI can use OCR to extract data, flag inconsistencies, and generate summaries for identity, business, and income documents, which are often the most time-consuming parts of a review.
  • Data retrieval and system silos: AI can help team members write their own SQL queries to pull data from warehouses, dramatically reducing dependency on data teams.
  • Product and feature proposals: AI tools can mock up full dashboard concepts and even generate code snippets, giving engineers clearer visuals and reducing friction between fraud and technical teams.

The power of empowerment and buy-in

  • Leadership plays a critical role in creating a culture where fraud fighters are encouraged to explore and innovate. 
  • The real value of time savings comes from how that freed time is reinvested, whether through reducing fraud losses, improving customer retention, or tackling new, high-impact initiatives. 
  • Teams are advised to keep proprietary and PII data out of AI tools and to find safe environments for experimentation, remembering that everyone is still learning what AI can and cannot do.
Full episode transcript
Share this episode