The End of Authorship? Generative AI and the Collapse of Open Access Norms
Generative AI has forced long-simmering questions about copyright, ownership, and authorship into the spotlight. Courts, regulators, and scholars are now grappling with a deceptively simple issue: who is the author of content created with AI? Is it the user who writes the prompt, the developer who built the model, the company that owns it—or no one at all?
Those questions matter. But they may not be the most important ones.
For the open access community, the deeper issue isn’t just ownership. It’s how generative AI challenges the assumptions that open access has relied on for decades: identifiable authors, clear rights, meaningful attribution, and stable systems of scholarly communication. In other words, AI doesn’t just complicate the rules—it reshapes the playing field.
The System Open Access Built
Open access was designed to remove barriers to knowledge. It emerged from a belief that research—especially publicly funded research—should be freely available to anyone, anywhere. Over time, institutions built repositories, funders adopted open policies, and researchers embraced Creative Commons licensing.
But even as it pushed for openness, the system retained a core principle: authorship matters.
Open access assumes there is a human creator who owns rights to a work and can decide how it is shared. Attribution isn’t optional—it’s foundational. It’s how scholars receive credit, build careers, and establish trust. Even when content is freely available, it is rarely anonymous.
That’s where generative AI creates tension.
What Happens When Authorship Gets Blurry?
AI doesn’t “author” in the human sense. It doesn’t intend, reflect, or exercise judgment. It generates outputs based on patterns learned from vast amounts of data. Humans may guide this process through prompts, but the connection between prompt and output can be surprisingly loose—especially when systems are complex or outputs are heavily transformed.
This creates a practical dilemma.
If authorship is unclear, so is everything built on top of it:
- Who owns the rights to an AI-generated article or dataset?
- Who chooses the license?
- Who gets credit in academic settings?
- Who is accountable if something is wrong, biased, or misleading?
These aren’t edge cases anymore. Researchers are already using AI to draft literature reviews, summarize findings, and generate visualizations. In some cases, AI is a tool. In others, it plays a much larger role.
Open access frameworks were not designed for this spectrum of involvement.
Attribution in a World Where “Nobody Writes”
Attribution has always served multiple purposes. It signals credibility, enables accountability, and ensures recognition. Open licenses, citation systems, and academic norms all depend on it.
But what happens when content isn’t clearly written by anyone?
Attributing AI-generated work to a human can exaggerate their contribution. Attributing it to a machine raises legal and conceptual problems—machines cannot hold rights. Leaving it unattributed removes important context for readers.
This isn’t just a legal puzzle—it’s an epistemological one. We assess knowledge based on who produced it: their expertise, methods, affiliations, and biases. When those signals disappear, so does part of what makes knowledge trustworthy.
In open science—where transparency and accountability are core values—this challenge becomes even more pronounced.
The Open Access Paradox
Here’s the paradox: open access helped create the conditions that enabled generative AI to flourish.
By making large volumes of scholarly content freely available, the open access movement expanded the pool of data that AI models could learn from. Open repositories, standardized metadata (like OAI-PMH), and accessible publications turned knowledge into something that could be mined, processed, and recombined at scale.
This is not a failure of open access—it’s a testament to its success. Greater accessibility has undeniably benefited science and society.
But it also means that open access contributed—directly and indirectly—to the rise of technologies that now challenge its foundations.
AI lowers the barriers to creating and distributing content, potentially advancing openness even further. At the same time, it destabilizes the norms—authorship, attribution, accountability—that made open access workable in the first place.
Where Do We Go From Here?
There is no single, obvious solution. But a few paths are beginning to emerge:
- Reinforce Human Authorship
One approach is to treat AI strictly as a tool. Under this model, human contribution remains central, and only humans can be recognized as authors. This aligns with how many legal systems currently interpret copyright. - Embrace Hybrid Models
Another approach acknowledges that AI can play a meaningful role in creation. Here, humans remain responsible for outputs, but disclosure becomes essential. Transparency about how AI was used could become a standard part of scholarly communication. - Accept Some “Authorless” Content
In some cases, AI-generated material may fall outside traditional copyright frameworks altogether. This could expand access—but at the cost of clarity around responsibility and governance.
Each pathway involves trade-offs. None fully resolves the tension between openness and accountability.
Beyond Copyright
It’s tempting to frame this debate as a question of ownership. But doing so risks missing the bigger picture.
Generative AI forces us to rethink how knowledge is created, validated, and shared in a world where content may not have a clear human origin. That’s a much broader challenge than copyright alone.
The open access movement has adapted to disruption before—from the rise of the internet to digital repositories to open science. AI is simply the next chapter.
The task now is not to resist that change, but to shape it.
Holding Onto What Matters
If there’s a lesson from the past two decades, it’s that technological change is inevitable—but values endure.
Openness, transparency, accountability, and trust were central to the original open access vision. They remain just as important today.
The difference is that preserving those values may require new tools, new norms, and new legal frameworks—ones that acknowledge a future where, at least sometimes, nobody writes.
And yet, knowledge still needs to be understood, trusted, and shared.
That part hasn’t changed.
Discover open access articles about AI and open access on the AGOSR database:
Generative Ai, Uk Copyright and Open Licences: Considerations for Uk Hei Copyright Advice Services
How Open Access Is Crucial to the Future of Science
Negotiating Open Access Ethical Positions and Perspectives
2 thoughts on “The End of Authorship? Generative AI and the Collapse of Open Access Norms”
My Experience with Artificial Intelligence: A Personal Perspective
Artificial Intelligence (AI) is one of the most remarkable technological developments of our time. From my personal experience, I believe AI is an extraordinary assistant that has the potential to transform education, research, engineering, and many other fields. However, its effectiveness depends greatly on the quality of the person using it.
One lesson I have learned is that to maximize the potential of AI, you must make it think the way you think. AI performs best when it understands your reasoning, writing style, problem-solving approach, and expectations. It is not merely about asking questions; it is about teaching the AI how you approach problems. The more clearly you communicate your thought process, the more effectively AI becomes an extension of your own intellect rather than a replacement for it.
Another observation is that AI processes information at a speed far beyond human cognitive ability. It often analyzes numerous possibilities before presenting a response that appears almost instantaneous. This rapid processing can give the impression that AI “thinks” ahead of us. In reality, it is drawing from vast amounts of knowledge and patterns while adapting to the context of our prompts.
I also believe that AI gradually builds an understanding of the user’s intellectual style during an interaction. It recognizes patterns in how questions are framed, how problems are approached, and what type of answers are preferred. As a result, the quality of its responses often improves as the conversation progresses. When a user consistently presents unique ways of analyzing or solving problems, AI adapts to that style, making the interaction increasingly productive.
For this reason, mastery of language is essential. Effective communication with AI requires precision, logical structure, and careful use of words. For Anglophone users especially, a strong command of English significantly enhances the quality of prompts and, consequently, the quality of AI-generated responses. AI is only as effective as the clarity of the instructions it receives.
I also hold the view that AI-generated academic work should still be regarded as an intellectual product when it reflects the original reasoning, experience, and analytical framework of its author. AI may assist in organizing ideas and refining language, but genuine scholarship comes from the human mind directing the process. Two individuals, even when using the same AI system, are unlikely to produce identical intellectual work because their experiences, reasoning patterns, perspectives, and objectives are inherently different.
Therefore, I see AI not as a substitute for human intelligence but as a powerful intellectual partner. Those who learn to guide it effectively will not diminish their thinking; rather, they will amplify it. The future belongs not simply to those who use AI, but to those who know how to make AI work in harmony with their own minds.
I wrote this sometimes ago, I also agree with your suggestion that some generative AI prompts should be more clearly defined by author and appropriately bounded, particularly those involving visual content.
Mahmud Dawud Mahmud
”Contribution of humans to production of knowledge will still remain relevant and critical even in the future.
AI and humans will work side by side, recognizing, accepting and embracing strengths and weaknesses of each side”.