Germany's AI Security Institute: Bitkom on Frontier Models

Germany plans to establish a security institute for artificial intelligence to catch up with countries like the UK, US, and France. Susanne Dehmel, a Bitkom executive board member, states that AI systems bring new risks, and this institute will help Germany assess them early and thoroughly. The new DE-AISI should have a research mandate focused on new "frontier models" and systemic risks to Germany's security. It should avoid areas like labor, consumer, and data protection, which are already handled elsewhere. Success depends on attracting top international talent, requiring funding comparable to the British AISI. The institute will focus on technical evaluation of advanced AI models and preparing security assessments for the government. This initiative is crucial for Germany's national security and sovereignty in the face of rapidly evolving AI capabilities.