GAISS 2026, IEEE Conference on Generative AI for Secure Systems

Research on using Generative models to model adversaries, enrich threat intelligence, and simulate attacks and defenses in realistic but controlled settings.

Attacks and defenses on Generative systems-prompt injection, jailbreaks, data and model poisoning, evasion, and certified or evaluated robustness under adversarial pressure.

Secure-by-design use of Generative AI in SDLC, CI/CD, and code generation-SAST/DAST augmentation, patch suggestion, supply-chain-aware tooling, and human-in-the-loop review.

Generative AI applied to industrial control, IoT, and cyber-physical systems: anomaly explanation, scenario generation, digital twins, and safety- and timing-aware deployment.

Intersection of Generative methods with quantum and post-quantum contexts: secure protocols, QML-assisted modeling, and analysis of threats and opportunities as quantum capabilities evolve.

Synthetic data generation, differential privacy, federated and distributed training of Generative models, and techniques that balance utility with confidentiality and compliance.

Generative approaches for secure protocols, network operations, SDN/NFV, and configuration: traffic analysis, policy synthesis, incident narration, and resilient infrastructure design.

How humans and Generative systems collaborate in security workflows; organizational, legal, and societal impacts; governance, documentation, and oversight of deployed AI.

Using Generative models to expand red-team coverage, discover attack surfaces, and automate defensive tasks-always with clear ethical boundaries and safe harbor for research.

Multi-step agents, tool use, memory, and autonomy in security-sensitive settings: containment, least privilege, monitoring, and evaluation of long-horizon agent behavior.