AI systems aren’t being hacked in traditional ways; they’re being manipulated, steered, and abused through the very interfaces designed to make them useful. In this session, a Bishop Fox cybersecurity expert walks through attack paths observed during hands-on security testing of AI-enabled systems and the models that power them in production. These aren’t speculative scenarios; they reflect how attackers move from input to impact by exploiting how AI systems interpret, connect, and act on data, revealing how small points of influence can translate into outsized risk across financial systems and sensitive data flows.

What you’ll learn:

• How AI model behavior can be influenced beyond intended controls
• How attacker access can persist across sessions, context, and system interactions
• Where boundaries between AI systems and connected data or services break
  down
• How these weaknesses translate into real business risk and where to prioritize
  controls

In financial services, these failures directly intersect with high-value systems: customer accounts, transaction workflows, and fraud controls. When AI is embedded in these processes, small points of influence can lead to unauthorized actions, data exposure, or breakdowns in controls that institutions rely on to manage risk and trust.

This session breaks down how those paths form and what it takes to prevent them, so you can better understand where your AI deployments are most exposed before attackers do.

Who should attend:

• CISOs and security leaders responsible for AI risk, governance, and investment
  decisions
• Risk, fraud, and data security leaders evaluating exposure in AI-driven processes
• Security and product leaders accountable for deploying AI safely at scale
• Architects and senior engineers responsible for designing and securing AI-enabled systems