RedStream is an AI security company founded by Tanner O'Donnell, who brings experience in AI security evaluation, terrorism studies, and open-source intelligence gathering. Currently in development, RedStream is building toward becoming a full-scale AI security platform with plans for expansion as the technology and market mature.
RedStream is being developed through a careful, iterative process that combines security expertise with technical innovation. As an early-stage company, our methodology continues to evolve as we refine our approach to AI threat detection and build toward a scalable platform
RedStream's testing framework has been informed by academic research in narrative manipulation, synthetic social media, and influence operations, particularly around simulation realism and adversarial behavior modeling.
Currently pursuing a Master's in Security and Terrorism Studies at the University of Maryland's START Consortium, where Tanner focuses on emerging technologies and extremist narratives. His undergraduate work at Hampshire College (2020) included research on how terrorist and violent extremist communities use digital platforms to escalate online activities to real-world harm, providing insights into digital threat pathways
Tanner participated in structured CBRNE red teaming exercises against frontier LLM systems as a counterterrorism specialist, testing high-risk misuse scenarios from an adversarial perspective alongside subject matter experts. He has also evaluated prototype LLM platforms from Palantir, IBM, and others as a test user in a Defense Innovation Unit initiative. In 2019, Tanner interned with the Syrian Archive and VFRAME, contributing to visual guides and training materials for machine learning tools identifying explosive remnants in conflict zones.
Through intelligence analysis work, Tanner gained experience with ICD-203 analytical writing standards and structured analytical products. This included weekly shifts as a watch officer in a GSOC environment, contributing to daily threat analysis reports and real-time threat monitoring - experience that directly informs RedStream's approach to identifying and tracking emerging AI misuse scenarios. He brings additional expertise in open-source intelligence gathering and extremist content analysis, with particular focus on monitoring active threat environments and adversarial narrative development.