AI-enabled features are now embedded in everyday software systems, from personalization and automation to analytics and decision support. While these systems promise speed and scale, many teams underestimate how ordinary engineering and product decisions around data, identity, and automation can quietly introduce security and trust risks.

This session examines real-world patterns of data misuse and trust failures that occurred not because of advanced attacks, but because of normal architectural choices made under delivery pressure. Using anonymized experience reports, the talk explores how AI-assisted systems amplify existing weaknesses in data handling, access boundaries, and accountability.

Rather than focusing on specific tools or vendors, the session emphasizes engineering judgment, responsibility, and ethical decision-making in modern software development. Attendees will gain a clearer mental model for reasoning about security and trust risks earlier in the lifecycle of AI-enabled systems.