We are proud to announce that Maksym Andriushchenko, Principal Investigator leading the AI Safety and Alignment group at the ELLIS Institute Tübingen, has co-founded Exponential Security Labs.
Co-founded alongside Naman Deep Singh, Anselm Paulus, and Prof. Dr. Matthias Hein, the Tübingen-based startup brings together top-tier technical expertise and research leadership to tackle one of the most critical frontiers in artificial intelligence: agentic AI security.
By combining their deep expertise across computer science, machine learning robustness, and AI safety, the team is uniquely positioned to build next-generation defenses for autonomous systems - exemplifying the strong synergy between the fundamental research at the ELLIS Institute Tübingen and the academic excellence of the University of Tübingen.
Securing the Next Generation of Autonomous AI
As AI agents become increasingly autonomous—making high-stakes decisions and automating complex workflows—ensuring their reliability, safety, and security remains an urgent, unsolved technical challenge. Exponential Security Labs aims to bridge this gap by securing agentic systems through self-improving red-teaming and guardrail agents, leveraging autonomous AI research.
The startup builds directly upon a decade of foundational research in adversarial robustness and AI safety—areas where Maksym has established a leading international reputation, including his work on prominent safety benchmarks like JailbreakBench and AgentHarm, as well as red-teaming contributions for frontier model developers.
Fostering the Tübingen AI Ecosystem
The launch of Exponential Security Labs highlights the vibrant translation pipeline within the Tübingen AI hub. By converting fundamental academic breakthroughs in machine learning robustness and alignment into robust industry applications, the founding team represents a vital step toward safeguarding modern AI infrastructure.
"Getting AI safety and alignment right is one of the most important challenges of our time," notes Maksym. "Translating our academic insights into real-world, scalable defenses through Exponential Security Labs is a natural extension of our mission to ensure advanced AI systems remain secure and trustworthy."
→ Exponential Security Labs website
→ Check Maksym’s research group here