Podcast 135: AI vs Fraud: Inside Device Fingerprinting

Our host, Igor Tomych, speaks with Catherine Woneis, VP of Product at Fingerprint, about how fraud detection works in a world where banking, commerce, and identity increasingly live inside browsers and mobile apps. Rather than theorizing regulation, the conversation focuses on the real mechanics of device intelligence, AI, and how trust is established digitally.

Woneis brings two decades of experience in analytics and fraud prevention, offering a practical view on how signals from devices and browsers are used to identify risk without turning fraud detection into user tracking. The episode explores where privacy boundaries should sit, and why intent matters more than any single data point.

Device Intelligence: Privacy, Consistency, and Trust

A central theme of the discussion is device fingerprinting and its frequent mischaracterization as surveillance. Woneis explains that modern device intelligence is designed to confirm consistency, not follow users across the internet. When someone accesses a financial service from the same device, configuration, and behavioral context, those signals help reinforce trust. When they change in unexpected or contradictory ways, they raise legitimate questions.

The conversation also addresses how privacy-focused platform changes by Apple and browser vendors affect fraud prevention. While these changes reduce invasive tracking, they also limit visibility for security teams. The challenge, as discussed, is finding a balance where privacy is respected while still enabling organizations to detect manipulation, automation, and misrepresentation.

AI, Automation, and the Human Weakness

The second half of the episode focuses on AI’s dual role in modern fraud. Fraudsters are using AI to scale attacks, generate realistic phishing attempts, and automate abuse, while financial institutions rely on machine learning to detect anomalies and assess intent rather than depend on static rules.

As agentic AI and automated browsing grow more common, Igor and Catherine highlight a key distinction: not all automation is malicious. The future of fraud prevention depends on differentiating legitimate agents acting on behalf of users from automated systems designed to exploit them. Despite these advances, both agree that humans remain the weakest link, with social engineering continuing to bypass technical safeguards. The episode closes with a reminder that digital trust requires layered defenses, context-aware signals, and continuous education, not just better technology.