German Parliament votes on national AI oversight law: Federal Network Agency becomes central authority for artificial intelligence supervision
What it really says
The German Bundestag is scheduled to debate and vote on the KI-MIG (AI Market Surveillance and Innovation Promotion Act, document 21/4594) on June 11, 2026. This law transposes the EU AI Regulation (EU 2024/1689) into German law and establishes the national supervisory structure for artificial intelligence. At its core, the Federal Network Agency (Bundesnetzagentur, BNetzA) becomes the central market surveillance authority, coordination point, and notifying authority for AI in Germany. A Coordination and Competency Center for the AI Regulation (KoKIVO) will be set up within the BNetzA to support all responsible authorities and ensure consistent legal interpretation. For particularly sensitive areas such as law enforcement and certain high-risk applications, an independent AI Market Surveillance Chamber will be established within the BNetzA. The bill also provides for AI regulatory sandboxes where high-risk AI systems can be tested in close cooperation with the supervisory authority. Sectoral responsibilities remain intact: BaFin oversees AI in the financial sector, BfArM handles medical devices, and data protection authorities address privacy-related violations. The Federal Cabinet approved the draft on February 11, 2026. The provisions of the AI Regulation generally apply from August 2, 2026.
Our assessment
For anyone worried about uncontrolled AI, this news is fundamentally reassuring: Germany is getting a clear supervisory structure for artificial intelligence. The KI-MIG is not a new set of rules but rather the necessary bridge between the already-applicable EU AI Regulation and German administrative reality. Without this law, no one would know exactly which authority is responsible for which AI violations from August 2026 onward. The choice of the Federal Network Agency as the central supervisory authority is pragmatic - it already has experience regulating digital markets and possesses the necessary technical expertise. At the same time, the approach is deliberately lean, with the ministry describing it as 'streamlined AI supervision.' This can be positive by reducing bureaucratic hurdles, but it also carries the risk that supervision may be understaffed and underfunded. The establishment of regulatory sandboxes is a smart compromise: companies can test innovative AI systems under supervision without immediately violating regulations. The fines follow the GDPR model and can be substantial - a signal that violations will not be treated lightly. Importantly, the law does not create new substantive rules but only enforcement structures. The actual prohibitions and obligations come directly from the EU regulation.
Relevance for Germany
For Germany, this law has immediate practical significance. From August 2026 onward, companies developing or deploying AI systems need to know whom to contact with questions and who will oversee them. The KI-MIG provides this clarity. This is particularly relevant for German SMEs (Mittelstand): many smaller companies are uncertain whether their AI use complies with the new rules. Regulatory sandboxes offer them the opportunity to test their systems under official guidance. For workers, the new supervisory structure means there will be a point of contact for complaints when AI systems are used questionably in the workplace, such as algorithmic performance evaluation or automated personnel decisions. However, the Federal Network Agency now needs to demonstrate it can handle this new task. It requires qualified personnel with AI expertise, which is highly sought after on the job market. Whether the allocated resources will be sufficient remains to be seen in practice. The vote comes at a time when the EU has simultaneously extended deadlines for high-risk AI through the Omnibus package, signaling that implementing AI regulation in Europe remains a balancing act between protection and innovation promotion.
Fact check
The primary source is the official German Bundestag website, confirming that the vote on the government bill (document 21/4594) is scheduled for June 11, 2026. The Federal Ministry for Digital and Transport (BMDS) documents the complete legislative process, including the Cabinet decision of February 11, 2026. The Bundesnetzagentur's role as central market surveillance authority is confirmed by both the BMDS and the legal publication activeMind.legal. The sectoral jurisdiction distribution (BaFin for finance, BfArM for medical devices) is consistently reported by multiple independent legal portals. The August 2, 2026 applicability date of the AI Regulation aligns with the timeline established in EU law, though the Omnibus package of May 2026 extended deadlines for certain high-risk applications to December 2027.
Source
- • https://www.bundestag.de/dokumente/textarchiv/2026/kw24-de-ki-1183820
- • https://bmds.bund.de/service/gesetzgebungsverfahren/gesetz-zur-durchfuehrung-der-ki-verordnung
- • https://www.ad-hoc-news.de/boerse/news/ueberblick/ki-mig-bundestag-startet-finale-fuer-deutsches-ki-gesetz/68972906
- • https://www.activemind.legal/de/guides/ki-mig/