AI and creative writing: Where does the tool end and where does authorship begin?

In January 2024, Japanese author Rie Kudan the prestigious Akutagawa Prize for her novel "Tokyo-to Dojo-to." In her acceptance speech, she explained that approximately five percent of the text was directly generated by a generative AI. These passages were not an oversight but rather part of the concept, as the novel itself deals with the soft, smoothing language of such systems.

This case illustrates how blurred the line has become. Anyone discussing creative writing today must clarify what a tool is allowed to do and when a text ceases to be one's own.

When the machine is recording

Language models have long since become part of the everyday lives of authors. They suggest synonyms, smooth out awkward transitions, or even provide an entire scene on request. This raises a practical question for literary competitions, publishers, and writing schools: how can a predominantly generated text be distinguished from one written independently?.

Some juries therefore run submitted manuscripts through Getsolved 's AI detector beforehand to mark passages with strikingly machine-like patterns and ask specific questions. However, such a notification does not replace a conversation with the person who submitted the text.

This is precisely where the limitations of such checks lie. A detector calculates probabilities; it doesn't judge originality. A very consistent, correct style can be just as noticeable as a genuinely generated paragraph, and someone writing in a foreign language is more likely to be suspected. Therefore, the result is useful as a reason for further inquiry, not as proof.

A spectrum instead of a line

The debate often suffers from the fact that "written with AI" can mean anything and nothing. A more helpful approach is to look at the individual work steps, as the machine's contribution increases from stage to stage.

The table shows that the real limitation lies not in the application of the technology, but in the control over the outcome. As long as a human decides what stays, what gets deleted, and why, AI remains a tool like a dictionary or a typewriter.

Literature has always played with rules and chance

The idea of ​​a writing apparatus is not new. Tristan Tzara suggested pulling newspaper words out of a hat and constructing a Dadaist poem from them. William S. Burroughs and Brion Gysin cut up texts and reassembled them in the late 1950s, while the Oulipo group wrote according to strict formal constraints.

In all cases, the process only provided material. The selection process created the text, and this same principle applies to language models.

What copyright law says about it

Legally, the situation is clearer than in the artistic debate. In both Germany and the USA, protection requires a human creation.

The situation in Germany

According to Section 2, Paragraph 2, copyright law only protects intellectual creations". A text generated by a machine without significant human input does not meet this requirement. " personal

However, those who select generated building blocks, modify them, and combine them into something of their own can certainly create a protected work. The crucial factor is how much personal expression is present in the final product.

The US position

the US Copyright Office published a comprehensive report following more than 10,000 public comments. The report states that prompts alone do not provide sufficient human oversight to make users the authors of the result. A distinction is made between AI as a tool that does not diminish copyright protection and AI as a replacement for human creativity.

Whether the control is sufficient will be examined on a case-by-case basis. Back in 2023, the agency had already protected the text and page layout of the graphic novel "Zarya of the Dawn," but not the images created with Midjourney .

What authorship actually entails

Beyond the legal paragraphs, an older question arises. Authorship isn't just about putting words on paper. It means making decisions and standing by them. A character dies because the author wants it that way. A sentence remains awkward because the author needs the friction. A model can simulate this intention, but it cannot take responsibility for it. Then there's the voice. It develops from experiences, readings, and idiosyncrasies that accumulate over years. A model trained on vast amounts of text, on the other hand, tends toward the middle, toward the most probable next word.

Those who write with AI can test themselves with a few questions.

• Could I justify every decision in the text, including stylistic ones?
• Did I modify the AI's suggestions or simply accept them?
• Would the text still work without the generated passages?
• Would I be willing to openly explain how the text came about?
• Does the result sound like my vote or like an average of all votes?

The more often the answer is an honest "yes", the more clearly the authorship lies with the human.

Transparency as the new norm

Rie Kudan publicly disclosed the amount of AI used in her work, and this has brought objectivity to the discussion. Many publishers and competitions now demand an explanation of whether and how AI was involved. This openness removes the suspicion of fraud and makes visible what would otherwise remain hidden.

This is beneficial for readers, as they can decide for themselves what value they place on a text. A brief mention in the afterword or acknowledgments is often sufficient, similar to how editing or translation are mentioned there.

Final thoughts

AI is changing creative writing, but it isn't eliminating authorship. It's simply shifting the focus of the actual work. Research, editing, and brainstorming can confidently be delegated. The selection, the approach, and the responsibility for what ultimately appears on the page remain human.

Where these three things are missing, the tool ends, it replaces the author, and then it is no longer creative writing, but production.

Joachim Rodriguez y Romero

Owner and Managing Director of Kunstplaza . Publicist, editor, and passionate blogger in the fields of art, design, and creativity since 2011. Graduated with a degree in web design from university (2008). Further developed creative techniques through courses in freehand drawing, expressive painting, and theatre/acting. Profound knowledge of the art market gained through years of journalistic research and numerous collaborations with key players and institutions in the arts and culture sector.

www.kunstplaza.de