Andreas is a PhD researcher at the Oxford Internet Institute focusing on human-machine teaming and human control of AI systems in computer network exploitation, defence and attack.
He finished his Master’s in 2020, where he researched AI-enabled sybil attacks and demonstrated how GPT-2 and Megatron-11b could be used against governments' public consultation websites to manipulate lawmakers and public opinion.
Currently, Andreas is developing the Algorithmic Resistance Cookbook, a guide to using data-driven tools and techniques to practice resistance against intrusive and repressive aspects of present-day algorithmic culture.
Andreas is the co-founder of ddocs.new and Fileverse.io which provide privacy-enhancing, peer-to-peer and end-to-end encrypted alternatives to Google Docs, Drive and Notion.
His most recent research articles include: The Cybersecurity Crisis of Artificial Intelligence: Natural Language-Based Attacks; Human control of AI systems: from Supervision to Teaming; and Artificial Intelligence for National Security: The Predictability Problem.
Mariarosaria Taddeo, Marta Ziosi, Andreas Tsamados, Luca Gilli and Shalini Kurapati, "Artificial Intelligence for National Security: The Predictability Problem," CETaS Research Reports (September 2022).
