CyLab announces Cyber Autonomy Initiative to advance AI-driven cyber defense

As artificial intelligence rapidly transforms both cyberattacks and cybersecurity defenses, Carnegie Mellon University’s CyLab Security and Privacy Institute has announced a new research initiative aimed at reshaping how digital systems protect themselves.

The CyLab Cyber Autonomy Initiative will bring together experts across computer science, engineering, and human-centered design to develop the scientific foundations for autonomous, AI-powered cyber defense.

Vyas Sekar, co-director of the Cyber Autonomy Initiative and Tan Family professor of Electrical and Computer Engineering, announced the initiative's launch today in San Francisco during the RSAC 2026 Conference. Sekar and nine other CyLab faculty members published a technical whitepaper outlining the research agenda for this initiative.

AI tools are making cyberattacks faster, cheaper, and more scalable, raising the stakes for organizations trying to protect critical infrastructure, businesses, and personal data. The Cyber Autonomy Initiative seeks to address this challenge by rethinking cybersecurity from the ground up, emphasizing automation, intelligent systems, and close collaboration between humans and machines.

“Attackers are already using AI to dramatically increase the speed and scale of cyber threats,” said Lujo Bauer, co-director of the Cyber Autonomy Initiative and professor of Electrical and Computer Engineering and Software and Societal Systems. “Our goal is to develop the foundational technologies that enable defenses to operate with comparable speed and sophistication, while still keeping humans meaningfully in the loop.”

The initiative’s research agenda is organized around four interconnected thrusts. The first focuses on building foundational AI and agent-based capabilities that enable autonomous attack and defense systems, including new abstractions, algorithms, and learning-based approaches. A second thrust examines how humans and AI systems can collaborate effectively, ensuring that autonomous tools are trustworthy, usable, and aligned with real-world operational needs.

A third thrust emphasizes the systems infrastructure required to support cyber autonomy, such as realistic emulation environments, scalable telemetry and analytics platforms, and secure testing and verification tools. The final thrust, known as the Cyber Autonomy Arena, will create competitive, system-level evaluations in which attackers and defenders face off in realistic network environments, helping researchers better understand how autonomous systems perform under real-world conditions.

“Cybersecurity has always been an arms race, but AI has dramatically accelerated the pace,” said Sekar. “By combining advances in AI, systems, and human-centered computing, we’re aiming to build a new generation of defenses that can adapt in real time and operate at machine speed, while still maintaining the accountability and safety of expert human operators.”

Through this interdisciplinary effort, CyLab aims to advance both the theory and practice of cyber autonomy while training students and researchers to tackle the next generation of security challenges. Ultimately, the initiative seeks to help organizations stay ahead of increasingly automated adversaries, ensuring that as AI reshapes cybersecurity, it does so in ways that are robust, responsible, and human-centered.

The Cyber Autonomy Initiative is currently seeking sponsors and research collaborators. For more information and to get involved, contact Michael Lisanti, CyLab senior director of partnerships, at 412-268-1870 or mlisanti@cmu.edu.