Frankfurt Draws a Line on AI: What the New Appeal Means for Authors

4 hours ago

5 min read

Frankfurt Book Fair opens today with artificial intelligence threaded through the professional programme, but one of the most important interventions has arrived outside the usual demonstrations of clever tools and faster workflows.

The German Publishers and Booksellers Association has issued what it calls the Frankfurt Appeal. Addressed to the European Commission, European Parliament and German federal government, it argues that generative-AI companies should disclose which specific copyrighted works have been used for training, respect rights reservations, and operate within a functioning licensing market rather than leaving authors and publishers to discover unauthorised use after the event.

For writers, this matters because it moves the AI argument away from the increasingly sterile question of whether we are 'for' or 'against' the technology. The real question is becoming much more practical: if human creative work contributes value to an AI system, what rights follow the work into that system?

The full demands are set out in the Frankfurt Appeal from the German Publishers and Booksellers Association.

The transparency problem comes first

Copyright is difficult to enforce when a rights holder cannot establish whether a protected work was used. That is the practical problem at the centre of the appeal. The association argues that AI companies have the technical ability to identify training material but do not provide work-level disclosure, leaving rights holders with litigation as one of the few routes to find out what happened.

That is a significant point for authors. A theoretical right is not much protection if exercising it requires an individual writer to take on an international technology company simply to discover whether their novel was included in a training dataset.

Transparency therefore comes before most of the other arguments. If a book can be identified, the author and publisher can discuss permission, licensing, payment and contractual control. If the book disappears into an opaque dataset, all of those conversations become harder.

Frankfurt is becoming an AI rights fair

The timing is not accidental. Frankfurter Buchmesse runs from 7 to 11 October this year, and AI, licensing, trust and content infrastructure are prominent across the professional programme. Sessions today include practical AI workflows for publishers and discussions about agentic AI, while later sessions address trust, licensing, content infrastructure and revenue from AI licensing.

The fair's official programme shows how deeply AI has entered publishing operations, from research and content workflows to discoverability and licensing. See the Frankfurter Buchmesse 2026 programme.

That combination is healthy. Publishing needs to understand what AI can genuinely do, not simply fear it. But the commercial conversation cannot be separated from the rights conversation. If publishers are exploring new AI revenue, the people who created the underlying books need clarity about where they stand.

Licensing is more interesting than prohibition

The strongest part of the emerging debate, for me, is not an attempt to stop AI. It is the possibility of building a legitimate market around access to high-quality human-created work.

We are already seeing evidence that such a market can exist. Publishers and technology businesses are experimenting with licensed access, retrieval systems and controlled use of content. Frankfurt itself is hosting discussions about unlocking revenue through AI licensing. That makes the claim that permission is simply incompatible with AI development increasingly difficult to sustain.

For authors, however, licensing raises another set of questions. Who owns the relevant rights under an existing publishing contract? Is AI training covered by clauses written before generative AI existed? Does the writer have approval rights? Can the author opt out? If money changes hands, what proportion belongs to the creator?

Those questions are not anti-technology. They are normal publishing questions. We already negotiate translation, audio, territorial, adaptation and other rights. If AI access becomes commercially valuable, authors should understand whether it is another right being licensed and who has authority to license it.

AI-assisted writing is not the same argument

There is also a distinction I think the industry needs to preserve. Debates about training rights are often collapsed into debates about authors using AI, as though they were the same thing.

They are not. I can use AI to help research a scientific idea, organise notes, test continuity, visualise a setting, prepare marketing material or manage the business surrounding my books without asking it to become the author. Those are tool uses. The judgement about what belongs in the story, the characters, voice, emotional choices and final prose can remain human.

AI-authored fiction is different. If a machine generates the substantive creative work and a person simply places their name on it, questions of authorship and reader transparency become unavoidable. Defending authors' rights over training data does not require us to pretend that every use of AI by a writer is illegitimate.

In fact, a mature position should be able to hold both ideas at once: authors can use powerful technology responsibly, and technology companies can be expected to respect the rights attached to human-created work.

What authors should watch next

The Frankfurt Appeal is not legislation. It is an industry demand aimed at lawmakers, and its proposals will be argued over. But it tells us where an influential part of European publishing wants the debate to move: identifiable training data, enforceable rights, meaningful licensing and consequences when rights reservations are ignored.

For working authors, the sensible response is not panic. It is attention. Keep publishing agreements accessible. Understand what rights you have granted. Watch for new AI clauses. Keep records of your creative process and original files. And when publishers announce AI partnerships, ask what content is being licensed and how authors participate.

The first phase of generative AI was dominated by capability: look what the machine can do. The next phase is increasingly about legitimacy: where did the material come from, who gave permission, who gets paid and who can say no?

Those may be less spectacular questions than a new model generating a novel in seconds. For authors, they are considerably more important.

About Rob Frankson

Rob Frankson is the author behind 121 Minutes and the Near Galaxy Saga, writing original science fiction while exploring how emerging technology is changing the practical business of being an author. About Rob Frankson

Join / Subscribe to 121 Minutes

AI & Editorial Transparency

AI & The Author is edited and published by Rob Frankson. Artificial intelligence is used to assist with news research, initial drafting, content organisation and supporting imagery. All articles are reviewed and, where necessary, edited by Rob Frankson before publication. The opinions, editorial position and final decision to publish remain the author's.