Guarding Against Malicious Biased Threats (GAMBiT)

The Guarding Against Malicious Biased Threats (GAMBiT) HSR1 dataset captures behavioral and technical data from a two-day cyber range experiment involving 19 red team participants. Designed to study attacker behavior and cognitive biases in realistic cyber attack scenarios, the experiment aimed to support the development of behavior classification methods. Participants were tasked with identifying valuable targets and exfiltrating data while operating in a simulated enterprise network environment built using the SimSpace Cyber Force Platform. The cyber range contained approximately 40 virtual devices and included embedded triggers to elicit cognitive biases such as loss aversion, base rate neglect, confirmation bias, and sunk cost fallacy. Data sources include network telemetry, keylogger data, terminal histories, and psychometric self-reports, including assessments of personality traits, decision-making competence, and risk propensity. The dataset supports hypotheses linking attacker performance and behavior to specific cognitive vulnerabilities and provides a rich foundation for research on human decision-making under adversarial conditions. GAMBiT HSR1 offers valuable insights into how cognitive factors influence cyber operations and can inform future training, detection, and defense strategies.