SAN FRANCISCO — OpenAI’s Cybersecurity Grant Program, launched in 2023, has quickly become a cornerstone for advancing research at the intersection of artificial intelligence and cybersecurity. The initiative, designed to equip cyber defenders with state-of-the-art AI tools, has attracted over 600 applications, highlighting the critical need for collaboration between AI innovators and the cybersecurity community.
Since its inception, the program has funded a diverse range of projects aimed at addressing pressing cybersecurity challenges. These include groundbreaking work from UC Berkeley’s Wagner Lab on defending against prompt-injection attacks in large language models (LLMs), and Coguard’s AI-driven solutions for reducing software misconfiguration, a common vulnerability in cybersecurity.
Mithril Security has developed a secure inference infrastructure for LLMs, ensuring that data remains protected even from system administrators, while Dartmouth’s Breuer Lab is pioneering techniques to safeguard neural networks against adversarial attacks without compromising their accuracy. Other notable projects include Boston University’s work on enhancing LLM capabilities to detect and fix code vulnerabilities, and UCSC’s research into autonomous cyber defense agents using foundation models.
OpenAI has also empowered the cybersecurity community by providing free access to ChatGPT Plus, a tool widely used by professionals to simplify technical language, write analytical code, and streamline incident response. This initiative is now expanding to include ChatGPT Team and Enterprise, starting with partnerships like the Research and Education Network for Uganda (RENU).
As OpenAI continues to drive innovation in AI-powered cybersecurity, it invites researchers and practitioners to submit proposals to join this growing community of innovators. The Cybersecurity Grant Program remains committed to enhancing defensive technologies and ensuring a secure AI-driven future.
