About the Initiative

Cornell University has launched the AI4AI initiative to address pressing security challenges in the rapidly evolving field of artificial intelligence.

Backed by a generous grant from Amazon, this groundbreaking effort brings together leading researchers from Cornell Bowers College of Computing and Information Science and Cornell Tech to develop safety protocols for AI agents, also known as agentic AI—an innovative technology capable of building software from simple user prompts.

“The launch of AI4AI at Cornell establishes a research home to address one of the most pressing problems in how we produce code today.” — Thorsten Joachims, Vice provost for AI Strategy, and director of the Cornell AI Initiative.

While agentic AI represents a revolutionary leap forward in software creation, it also carries significant risks. Left unchecked, these AI systems can generate incorrect, misleading, or even malicious code, leaving room for exploitation by hackers through attacks like prompt injections and unsafe executions.

The AI4AI initiative seeks to mitigate these vulnerabilities by creating a robust security framework, complete with rules and verification processes, to ensure agentic AI delivers secure and reliable outcomes.

Led by Alexandra Silva, professor of computer science at Cornell Bowers, and Vitaly Shmatikov, professor of computer science at Cornell Tech and Cornell Bowers, the initiative leverages Cornell’s deep expertise in machine learning, programming languages, and system verification. “Agentic software represents an entirely new computing paradigm, but it doesn’t come without risks,” the researchers explained. The AI4AI team aims to unlock the transformative potential of agentic AI while safeguarding its integrity and public trust.