AI disclosure requirements in Germany: what applies from 2 August 2026

The EU AI Act's transparency obligations are a Union-wide regulation, so the substance is the same in Germany as everywhere in the EU: from 2 August 2026, AI systems that interact with people must disclose they're AI, synthetic media must carry machine-readable marking, and deepfakes and certain AI-generated public-interest text must be labeled. What differs country by country is who enforces it, in what language your users encounter it, and how quickly complaints turn into questions — which is what this page covers for Germany.

Enforcement in Germany

Germany is implementing the AI Act through the KI-Marktüberwachungs- und Innovationsförderungsgesetz (KI-MIG). The Federal Cabinet adopted the government draft on 10 February 2026 and the bill is now in the parliamentary procedure (Bundestag and Bundesrat), after Germany missed the Act's 2 August 2025 deadline for designating national authorities. Under the draft, the Bundesnetzagentur (BNetzA) becomes the central market surveillance authority — including for Article 50 transparency duties — with sector regulators such as BaFin keeping jurisdiction in their own domains, and a KoKIVO coordination centre inside BNetzA pooling cross-authority AI expertise. Until the KI-MIG is enacted the supervisory map is draft, not final — but Article 50 itself applies from 2 August 2026 regardless, with the Act's penalty ceiling of €15M or 3% of worldwide turnover.

The four duties, as they show up on your site

• Article 50(1) — chatbot disclosure. Any AI system interacting with visitors must make that clear no later than the first interaction, in a clear and distinguishable way. A line in your terms doesn't satisfy it.
• Article 50(2) — machine-readable marking. AI-generated audio, images, video, and text must be marked in a machine-readable, detectable format. For generative systems already on the market before 2 Aug 2026, this specific duty has a transition until 2 Dec 2026 — the rest does not.
• Article 50(3) — emotion recognition & biometric categorisation. People exposed must be informed the system is operating.
• Article 50(4) — deepfake and AI-text labels. Visible disclosure for deepfakes, and for AI-generated or manipulated text published to inform the public on matters of public interest.

Germany-specific practicalities

• BNetzA already runs market surveillance for telecoms and radio equipment — expect process-driven, documentation-heavy supervision once the KI-MIG lands, the style German regulators are known for
• Sector overlap is real: an AI feature inside a financial product answers to BaFin, not BNetzA — map which regulator owns your sector before assuming the central authority
• German consumer-protection and competition associations (Verbraucherzentralen, Wettbewerbszentrale) can pursue unfair-competition claims over misleading practices — an undisclosed AI chatbot is exactly the kind of gap they test, independent of the AI Act's own fines
• Keep your evidence file ready in German: screenshots of the German-language first-interaction state, plus your widget settings export with a date

Disclosure language should match your user-facing language. For a site serving Germany in German, a first-message line such as "Sie chatten mit einem KI-Assistenten" (you're chatting with an AI assistant) is the pattern to aim for — clear, at first interaction, not buried in an Impressum or ToS.

Common questions