ORNL Unveils Photon to Secure AI at Exascale

Researchers at Oak Ridge National Laboratory (ORNL) launched "Photon," a new framework designed to find AI security flaws ^[1]. The system uses the Frontier exascale supercomputer to test AI models against many attack scenarios at once ^[2]. This process helps experts identify risks in critical areas like healthcare and national security before they are exploited ^[1][2]. Photon allows for rapid detection of adversarial exploits by scaling testing across thousands of computing nodes ^[3]. Previously, these tests required significant time and resources to complete ^[2]. By automating vulnerability discovery at this scale, the team aims to make AI systems more robust for public and private sector use ^[1][3].