Oregon State University is a public research university in Corvallis, Oregon, whose hardware security research group contributed to AI compute governance through the Survival and Flourishing Fund's FlexHEG (Flexible Hardware-Enabled Guarantees) program.
Oregon State University is a public research university in Corvallis, Oregon, whose hardware security research group contributed to AI compute governance through the Survival and Flourishing Fund's FlexHEG (Flexible Hardware-Enabled Guarantees) program.
People– no linked people
Updated 05/18/26Funding Details
Updated 05/18/26- Annual Budget
- $1,850,000,000
- Current Runway
- -
- Funding Goal
- -
- Funding Raised to Date
- -
Org Details
Updated 05/18/26Oregon State University (OSU) is a public land-grant research university located in Corvallis, Oregon, founded in 1868. It is the largest university in Oregon, with approximately 38,460 students and campuses in Corvallis, Bend (OSU-Cascades), Portland, Newport (Hatfield Marine Science Center), and a large online program (Ecampus). OSU had record research expenditures of $422 million in fiscal year 2024, and its endowment surpassed $1 billion in 2025. OSU was the first institution in the United States to offer both Master's and Ph.D. degrees specifically in Artificial Intelligence. The university has deep research strengths in AI, machine learning, robotics, cybersecurity, and semiconductor engineering. Notable AI-related research infrastructure includes the Collaborative Robotics and Intelligent Systems (CoRIS) Institute, directed by Kagan Tumer, which advances theory, design, and deployment of robots and intelligent systems with more than 45 faculty and 180 graduate students. In the AI safety domain, OSU has several relevant research threads. Tom Dietterich, Distinguished Professor Emeritus and a pioneer of machine learning, has long worked on safe artificial intelligence including safe reinforcement learning, open category supervised learning, and explainable AI. His research has been supported by DARPA, NSF, and the Future of Life Institute. Sandhya Saisubramanian, an assistant professor, leads the Intelligent and Reliable Autonomous Systems (IRAS) group focusing on safe, reliable, and unbiased AI decision-making. Most directly relevant to x-risk is the work of Vincent Immler, an assistant professor of electrical and computer engineering who specializes in hardware security, anti-tamper mechanisms, physical unclonable functions (PUFs), and secure processor design. Immler, who holds a PhD in hardware security from TU Munich and previously worked in cryptanalysis for the German government, is a co-author on the paper 'Hardware-Enabled Mechanisms for Verifying Responsible AI Development.' OSU received a $271,000 grant from the SFF-2024 FlexHEG round (funded by the Future of Life Institute and Blake Borgeson) for work on flexible hardware-enabled guarantees, which are mechanisms for AI accelerators that enable privacy-preserving verification of compliance with policies governing AI development and use. This work aims to develop tamper-resistant hardware that can enforce agreed-upon rules about AI training compute, supporting international AI governance frameworks.
Theory of Change
Updated 05/18/26Oregon State University's AI safety-relevant work, particularly through the FlexHEG grant, operates on the theory that hardware-level enforcement mechanisms can provide trustworthy, tamper-resistant verification that AI developers are complying with agreed-upon policies regarding compute usage, training scale, and deployment. By developing physical tamper protection and secure processors that monitor AI accelerator usage, this approach aims to create the technical infrastructure needed for effective international AI governance agreements. If powerful AI systems pose existential risks, having reliable hardware-based mechanisms to verify and enforce development constraints could be a critical layer of defense that complements software-level safety measures and policy frameworks.
Grants Received
Updated 05/18/26Projects– no linked projects
Updated 05/18/26Discussion
No comments yet. Be the first to share your thoughts.