AI Training Opt-Outs Now Carry Real Penalties in the EU. pdfRest Puts the Opt-Out Inside the PDF Itself.

Courts have ruled that terms-of-use language doesn't qualify as a reservation. Only an embedded, machine-readable signal like TDMRep does.

CHICAGO, IL, UNITED STATES, August 26, 2026 /EINPresswire.com/ -- The grace period is over. The EU AI Act's copyright obligations for providers of general-purpose AI models, including the requirement to identify and respect rights reservations made under the EU text-and-data-mining exception through machine-readable means, have applied since August 2, 2025. One year later, on August 2, 2026, the enforcement provisions took effect, exposing providers to fines of up to 15 million euros or 3 percent of global turnover. Even as Brussels debates delaying other parts of the Act under the proposed Digital Omnibus, the copyright obligations have held their schedule. The obligation is no longer new, and non-compliance is no longer free

The legal standard for those reservations is already settled in Germany. In December 2025, the Hamburg Higher Regional Court held in Kneschke v. LAION that a text-and-data-mining opt-out is effective only if it is machine-readable. A notice in the terms of use is not enough. Among the methods the court recognized: robots.txt, HTTP headers, and the TDM Reservation Protocol (TDMRep) developed by the W3C.

There is a gap in that list that matters to anyone who publishes documents rather than webpages. Robots.txt and HTTP headers protect content at the website level. PDF documents do not stay on websites. They are downloaded, emailed, re-hosted, mirrored, and passed through document pipelines, and the moment a file leaves the domain it was published on, the website-level opt-out protecting it stays behind.

pdfRest, the PDF processing API toolkit from Datalogics, closes that gap with its TDM Reserve PDF endpoint: a single API call that embeds a TDM Reservation Protocol policy, designed to support the W3C TDM Reservation Protocol, this tool enables content owners to clearly state their Text and Data Mining (TDM) policy, ensuring compliance with the EU CDSM Directive Article 4 and providing a standardized signal to AI crawlers and scrapers. The reservation becomes part of the file. Wherever the document goes, the machine-readable rights signal goes with it.

"The law now recognizes machine-readable reservations, and courts have said plainly that a sentence in your terms of use is not one," said Eric Shore, Chief Innovation Officer at Datalogics. "But every recognized mechanism assumed content lives at a URL. Documents move. Embedding the reservation in the file itself is the only way the signal survives the journey, and for organizations whose intellectual property ships as PDFs, that is the difference between an opt-out that exists and one that works."

The endpoint is built for scale: publishers, research organizations, standards bodies, legal teams, and any business whose documents constitute intellectual property can apply reservations across entire document libraries programmatically, as part of the same pipelines that already produce, convert, and archive their files. TDM Reserve PDF is available through pdfRest's Cloud, Container, and AWS deployments, and like every pdfRest operation, it is built on the Adobe PDF Library, the same technology behind Adobe Acrobat.

The reservation expresses the rights holder's position in the machine-readable form that the EU framework contemplates.

Organizations can apply a TDM reservation to a document in minutes using API Lab or review additional documentation regarding PDF security.

About pdfRest

pdfRest is the PDF processing API toolkit from Datalogics, offering more than 50 document operations through a developer-friendly REST API. Built on the Adobe PDF Library and available as Cloud, Container, and AWS deployments, pdfRest lets development teams convert, secure, extract, optimize, and archive PDF documents at scale, on infrastructure they control. Learn more at pdfrest.com .

Eric Shore
Datalogics, Inc.
+1 480-221-1467
email us here