---
title: "Authorship in the AI Era"
description: "When we encounter an original work, we generally assume that we know who authored it. However, with the development of artificial intelligence, we are having…"
url: https://www.independentpress.com/article/authorship-in-the-ai-era
date: 2026-09-12
categories: ["Artificial Intelligence"]
author: "Bailin Lu"
---

# Authorship in the AI Era

![photo-1591696331111-ef9586a5b17a](https://images.ctfassets.net/ewtdlsoyixc1/1zoWSEalSrs0Q3Xd9CsXUv/132be9d95da7f5e45b011fb2d892e82e/photo-1591696331111-ef9586a5b17a.avif)

When we read an article or a book, watch a film, or listen to a song, we generally assume that we know who authored it. For decades, a systematic framework has been in place to attribute credit, both legally and artistically, to the creator or owner of an original work.

In both legal and industry contexts, we have adopted a system that can effectively ensure that responsibility and rewards are attributed to the right contributors. However, with the development of generative artificial intelligence, we are having to rethink the concepts of creativity, intent, ownership, and what it means to be an author in the digital age. A person may supply an idea while a large language model (LLM) produces the wording. A person may revise an argument generated by a prompt they entered into an AI agent. In some cases, an individual may simply publish output generated by AI that they did little to nothing to shape. Currently, there is no generally accepted descriptive credit or disclosure that tells us which of these processes the author performed.

Generative AI’s influence on the creation of original work raises important questions about what it means to claim authorship in the modern era. In the age of AI, much of the debate centers on whether AI itself can be an author. However, answering that question alone does not settle the issue because a more pressing question remains:

**Should a person using an AI agent be considered the author of everything it generates?**

A simple example can make the issue clear: suppose that two people are submitting articles to the same publication. The first person spends days researching a subject, developing an argument, and writing a draft. Eventually, they ask an AI agent to correct grammatical mistakes. Meanwhile, the other individual provides the system with a topic and a few instructions and asks it to write an article. Both then submit their output without changing a word. When submitting their articles, both attach the same disclosure:

_"AI was used in the preparation of this article”_

The statement is true in both cases, but it conceals almost everything we need to know about their respective contributions as authors. One author supplied the argument and its expression; the other simply supplied a request and hit enter. These two cases are rather straightforward and represent easily defined and common uses of AI that occur in the creative processes. If we put them at two ends of a spectrum, between them lie many more ambiguous cases such as people who use AI to develop ideas, organize their own thoughts, generate passages, or to revise what they have already written.

How should we describe authorship across these two different processes? Knowing whether AI was involved is only the beginning of the necessary disclosure, but it is not sufficient to attribute AI’s contribution to the creative process and establish authorship. What we also need to understand is what the person claiming authorship precisely created or contributed. Let’s first look at the definition of what it means to be an author.

In philosopher Risto Hilpinen’s paper [_On Artifacts and Works of Art_](https://doi.org/10.1111/j.1755-2567.1992.tb01155.x), he connects authorship to the intentions through which something is created:

_“The existence and some of the properties of an artifact depend on an author’s intention to make an object of [a] certain kind.”_ (Hilpinen, 1992, p. 65)

An artifact, by definition, is something intentionally made, including a physical object or an intellectual work such as a written text. Authorship therefore concerns a relationship between what someone intended to create and what their activity brought into existence. This relationship helps explain why involvement alone is insufficient to establish authorship of an entire work. For example, printing a novel does not make the printer the author of its story. Equally, writing the story does not make the novelist the author of its cover design. Different aspects of the same book could arise from different people’s intentions. That is, we need to establish what kind of contribution, or intention, makes someone an author of an artifact.

Building on this account of authorship, we can first propose a baseline: 

**To qualify as an author of an artifact, a person must contribute a creative intention that is realized in its content or expression.**

Take literary articles as an example. A person must help determine what the article says or how it says it to be considered an author of an article. This may involve developing the article’s main argument, constructing the dominant explanations within the text, or composing language through which the article’s ideas are conveyed. Creativity, understood as an unprecedented discovery or invention, is not strictly required. What is required is a contribution that goes beyond simply reproducing or executing someone else’s decision.

On the account adopted here, generative AI does not possess such intentions, since its output is not an intentional undertaking of its own, even when the output goes beyond what the user specified. Under this account, AI therefore does not qualify as an author.

However, excluding AI from the discussion of authorship does not mean that its user automatically becomes the author of everything it generates simply by typing a prompt into an LLM and hitting return. This is because the human must still satisfy the same requirement: their creative intentions must account for the aspects of the article they claim to have authored. Some generated elements may consequently remain without an author, rather than becoming the users by default.

This brings us back to the original question: for an AI-assisted project, what can the human who prompted the generative output legitimately claim?

The way we ordinarily assign credit is already a compromise between accuracy and practicality. A byline names the writer, but an article may draw on arguments developed by previous thinkers, concepts inherited from earlier scholars, and expressions that existed long before the writer began writing. Those contributions have histories of their own. However, if we try to trace them all, authorship would take the form of a system resembling a family tree, with nearly endless links that trace all the way back to its theoretical origins [(Lu, 2025, Chapter 3)](https://scholarship.miami.edu/esploro/outputs/graduate/The-Author-is-Not-Dead-Just/991032728721802976). That is, a complete account would be practically impossible to assemble, let alone fit beneath a title.

In daily practice, what we recognize as authorship is therefore partial: calling someone the author of an article does not mean they originated everything within it. We recognize a particular contribution while leaving much of the work’s creative history unstated. The difficulty is that this useful shorthand can also conceal claims that extend beyond what a person specifically created. The recognition of partial authorship makes the limits explicit. For instance, when assigning credit for an article, a claim of authorship should be accompanied by a description of the individual’s specific contributions. Someone might have developed the argument, written its explanation, or substantially revised its reasoning. Describing these roles allows us to acknowledge authorship without pretending that every contribution belongs to the name on the page.

We can call this system of attributing credit “**descriptive credit**.”

Adopting the practice of attributing descriptive credit can solve two primary problems:

It can explicitly clarify the roles in a project involving multiple contributors without having to quantify different kinds of contributions and fight over whose name goes first.

It can help clarify disputed authorship claims by distinguishing the artist’s conceptual contribution from the contributions of the object’s original designers and makers.

More importantly, it provides us with a clear pathway to regulating authorship attribution in AI-assisted work.

A disclosure that states only whether AI was used leaves these differences unexplained. Given the rapid pace of development in the artificial intelligence industry, a future in which AI becomes inseparable from daily work appears imminent. To prepare for that world, it is now a pressing issue to design a system that better regulates how AI is employed in the context of original authorship.

One possible solution is “**descriptive disclosure**.”

Similar to the proposed use of descriptive credit, a descriptive disclosure of AI usage provides information that is far more practical in gauging authorship of original work. It can clarify how AI was used, what AI was used, what AI generated, what the user provided, and what the user subsequently revised. Such descriptions would give audiences a better basis for understanding what accredited authorship truly denotes.

Let us return to our previous example of two people submitting articles with AI disclosures. The first person’s claim to authorship rests on the argument and writing they produced. The second person’s prompt initiated a process, but it did not supply the creative contribution needed to justify the same credit. Separating descriptive credit from descriptive disclosure makes these differences explicit.

Under the current proposals, the first person would include:

**Descriptive Credit:** Research, argument development, and writing by [Name].

**Descriptive Disclosure:** AI was used solely to correct grammatical errors.

Under the current proposals, the second person would include:

**Descriptive Credit:** Topic selection and general instructions by [Name].

**Descriptive Disclosure:** AI generated the article’s argument, structure, and wording in response to the supplied topic and instructions. The output was submitted without revision.

Both people receive credit for what they contributed, while the separate disclosure explains the role AI played. Crediting the second person’s involvement does not require presenting them as the article’s author.

Of course, putting this system into practice would require further discussion and planning. Questions remain about how detailed disclosures should be, how contributions can be verified, and how to describe cases in which human and AI inputs become difficult to separate. Different fields may also need different standards, and any disclosure requirement must remain manageable for both contributors and publishers. Descriptive credit and disclosure should therefore be understood as a proposed direction for improving attribution in the AI era, with the practical aim of providing enough information to make authorship claims meaningful without turning every creative project into an exhaustive record of its production.

**Descriptive Credit:** Philosophical framework and underlying research by Bailin Lu, developed in his master’s thesis, _The Author Is Not Dead, Just Distributed_. Article direction, argument refinement, and final editorial review by Bailin Lu.

**AI Usage Disclosure:** AI assisted with outlining, revising passages, discussing the argument, correcting grammar, and locating and formatting citations. The author guided the process and reviewed the final text.