International Edition
Latest News
Entertainment

Hamburg Court Rules Natural-Language Terms Can Block AI Training Data

Hamburg court rules natural-language terms may block AI training data A Hamburg regional court decided that a natural-language reservation in a website’s terms of service can function as a machine-readable opt-out under German copyright law, potentially expanding the…

Hamburg court rules natural-language terms may block AI training data

A Hamburg regional court decided that a natural-language reservation in a website’s terms of service can function as a machine-readable opt-out under German copyright law, potentially expanding the ways rights holders can block commercial text and data mining.

The ruling, issued Sept. 27 in a case brought by the LAION association, interprets § 44b of the German Copyright Act (UrhG), which permits reproductions of lawfully accessible works for text and data mining unless the rights holder has reserved use in a machine-readable form. The court found that a reservation written in natural language within the terms of use may satisfy that requirement, itmr-legal.de reported. The decision addresses a gap that has persisted since the EU’s 2019 Copyright Directive required machine-readable opt-outs but left the technical standard undefined. While robots.txt files and metadata tags have been treated as the default compliance method, the Hamburg court held that a clearly worded contractual clause can also be “machine-readable” if automated systems can reliably detect and parse it. LAION, a non-profit that assembles large-scale image-text datasets used to train models such as Stable Diffusion, had argued that only technical standards like robots.txt qualify. The court disagreed, suggesting that the obligation falls on the miner to build tools capable of reading standard legal notices. The ruling is a first-instance decision and subject to appeal.

Section 44b UrhG implements Article 4 of the DSM Copyright Directive for commercial text and data mining. A separate research exception in § 60d UrhG allows non-commercial scientific research to bypass opt-outs entirely. Providers of general-purpose AI models also face separate transparency and copyright-compliance obligations under Article 53 of the EU AI Act. FAQ

What did the Hamburg court decide? The court ruled that a natural-language opt-out clause in a website’s terms of service can meet the machine-readability requirement of § 44b UrhG, allowing rights holders to reserve their rights without technical files such as robots.txt.

Does this ruling apply to all AI training in Germany? It is a first-instance judgment in a single case (310 O 227/23) and can be appealed. Other courts are not bound by it, but it signals how at least one court interprets the machine-readability standard. What about scientific research?

Section 60d UrhG provides a broader exception for non-commercial scientific research that is not overridden by a rights holder’s reservation. Are AI model providers affected? Yes. Under Article 53 of the EU AI Act, providers of general-purpose AI models must implement a copyright-compliance strategy that respects valid opt-outs and publish a summary of training content.

About the author: Entertainment Desk

Entertainment Desk at Archynewsy.