Skip to main content
EnglishEnglish
Digital Skills and Jobs Platform
AISI Germany: New Institute for Artificial Intelligence Security

AISI Germany: New Institute for Artificial Intelligence Security

With the AI Safety and Security Institute Germany (AISI Germany), Germany has created a national competence and evaluation institution for high-performance AI systems. The Institute started its work in Berlin on 31 August 2026, following a decision of the National Security Council in June 2026.

With the AI Safety and Security Institute Germany (AISI Germany), Germany has created a national competence and evaluation institution for high-performance AI systems. The Institute started its work in Berlin on 31 August 2026, following a decision of the National Security Council in June 2026. It was set up jointly by the Federal Ministry for Digital and State Modernisation (BMDS) and the Federal Ministry of the Interior (BMI).

The background is the growing capabilities of modern AI models and the security risks that come with them. In areas such as cyber operations, software development and automated decision-making, new requirements are emerging for the technical evaluation of AI systems. Other concerns include digital sovereignty and dependence on non-European technology providers, the protection of critical infrastructure and the implementation of the EU AI Act.

Roles and responsibilities

AISI Germany is to examine high-performance AI models technically and assess their security-relevant effects on the state, the economy and society. Its tasks include analysing cyber capabilities, levels of autonomy, robustness and potential for misuse, as well as monitoring technological developments internationally. It also aims to build national expertise in AI security.

On this basis, the institute advises the Federal Government on AI-related security and sovereignty issues, development scenarios and strategic needs for action. It is explicitly not a regulatory or market surveillance authority, but an analytical and operational competence centre.

Cooperation between BSI and the Federal Network Agency

The institute is being built up in two phases. In the first phase, two federal authorities form its "virtual" core: the Federal Office for Information Security (BSI) and the Federal Network Agency (BNetzA). The two ministries have entrusted them with the corresponding tasks.

The BSI, Germany’s central cybersecurity authority, covers the “security” dimension, focusing on risks from cyberattacks and the misuse of AI systems. The Federal Network Agency covers the “safety” dimension, which includes risks that may arise from the functioning and behaviour of AI systems, in particular their reliability and controllability.

Further expansion and international cooperation

In the second phase, the institute’s technical and staff capacity is to be expanded. It is also meant to ensure a structured transfer of security-relevant knowledge on AI between the Federal Government, the administration, business and civil society. For now, the focus is on advising politics and administration, although offers for a broader audience are already being planned. The institute also publishes a monthly overview of current AI security topics (in German).

Internationally, according to the BMDS, technical exchange is already taking place with partner institutions in France and the United Kingdom and with the EU AI Office. The institute is also expected to take part in the international network of AI safety institutes and to contribute to common standards for the safe handling of advanced AI systems.

News details

Geographic scope - Country
Germany
Geographical sphere
National initiative