Over-trust is excessive confidence in a system’s output or behavior, even when its limits are not clear. In cyber security, it happens when people treat a tool as more accurate, safer, or more authoritative than it really is. This is especially risky with AI systems, automated alerts, and chatbots, where fluent responses can hide mistakes, bias, or missing context.
Attackers can benefit from over-trust by making users rely on bad advice, approve unsafe actions, or disclose sensitive data to a system that seems friendly and dependable. Defenders reduce this risk with clear boundaries, scoped permissions, human review for important decisions, and user training that emphasizes verification. In secure design, the goal is to make a system useful without encouraging blind confidence in its answers.