Germany's AI Security Institute: Assessing AI Risks

Germany is building a national AI Security Institute to assess advanced AI models and their potential security risks. This new institute, called DE-AISI, began operating as a virtual nucleus in early August. It draws on expertise from the Federal Office for Information Security and the Federal Network Agency. The decision to establish DE-AISI was adopted by Germany’s National Security Council. Technical exchanges are also underway with partner institutions in France, the UK, and the EU AI Office. The institute aims to provide scientific and technical expertise to support government risk assessments. It is designed to complement the EU AI Act, not act as another enforcement body. The government is currently using a step-by-step, virtual model while decisions are made on its permanent structure, funding, and location. This development is important for understanding how governments plan to manage the evolving risks of artificial intelligence.