If You Don’t Disclose AI, You’re a Plagiarist

On January 27^th, the British Museum published four images to its Instagram account encouraging visitors to take a closer look at their various exhibits. 

However, several followers of the account felt that something was amiss the images. They wondered if the images were generated by AI or if the account had been hacked. 

Louise Archeology was one of the many who raised concerns about the image (video version). She was unfollowed by the museum, and the images came down without an immediate statement from the organization about what happened. 

Though there is no way to 100% confirm these images were AI, Louise tracked the tags in the post to an AI persona and a company named V8 Global, which specializes in AI marketing. 

According to an article by Jo Lawson-Tancred at Artnet, the images were “user-generated” and that the museum does not “post AI-created images,” thus why they were deleted. However, the museum did not explain how the images made it through their review process. 

Regardless of how the images were published, it seems almost certain that the British Museum published either AI-generated or AI-enhanced images to its account and failed to clearly disclose that it was doing so. 

As Louise pointed out, the British Museum already has a complex history with many of the antiquities it houses. Adding AI into that mix is just throwing gasoline on a fire.  

But, if it is true that the museum published AI-generated images knowingly and without clear disclosure, I would argue that they committed plagiarism. I do not mean that in an expanded sense. I mean it in the straightforward one. The museum took credit for work that it did not do. 

That, in turn, is a big part of why so many are upset about it. Not just that AI was used, but that it was used deceptively. 

The Nuances of Plagiarism 

Plagiarism is an incredibly complicated subject because there is no singular standard of what is and is not plagiarism. 

Plagiarism in academia is different from in novel writing, which is different from in journalism, which is different from a social media post. Different kinds of work have different citation standards and it’s important to code switch between them to avoid running afoul of the existing standards. 

In the end, it comes down to the audience’s expectations.  A celebrity “writing” an autobiography can use a ghostwriter with little backlash. The audience does not expect that person to author their own book. But a creative novelist will face severe backlash if they use a ghostwriter because the expectations are different.  

However, even that is not an absolute truth. Carolyn Keene, the “author” of the Nancy Drew stories, is a pseudonym that has been carried by dozens of ghostwriters. But, once again, the audience knows this and accepts this. The name serves more as a brand than a claim of authorship. 

This brings us to AI. As ubiquitous as the technology has become, the broad expectation on the internet is that works shared online, unless clearly marked as AI, were generated by humans.  

This includes your photos on social media, articles published on Medium or Substack, and your videos posted on YouTube. The default expectation is that the content we access and encounter was created by human hands driven by a human mind. 

The reason for that is simple. One of the primary reasons, if not, the primary reason we use the internet is to engage with and interact with other humans. You are reading this article now and whether you agree with me or disagree with me, you are engaging with the thoughts, actions, and creativity of another person. Me.  

AI robs us of that connection. By its very name, it is artificial. When you present something artificial as authentic, especially when you put your name to it, you are taking credit for creative decisions you did not make, work you did not do, and ideas you did not generate. 

To my mind, that is the very essence of plagiarism. Just because there is no human victim being plagiarized doesn’t mean that there aren’t other victims, namely the deceived audience.  

And, as the British Museum case shows, is a very real problem with real-world consequences. 

AI is Not “Just Another Tool” 

Ever since the public launch of ChatGPT, many have tried to argue that AI is “just another tool,” claiming it is just a more advanced calculator or grammar checker. But that is not the case.  

AI is often wrong. Relying on AI when you do not understand the underlying concepts of what you are trying to achieve is dangerous. Ask lawyers who submitted pleadings with fake citations generated by AI. Look at what happens to those who “vibe code” and create serious security issues.  

That is because AI is not just another tool. It is making creative and intellectual decisions on behalf of its operators. No matter how specific your prompt gets, the AI system is making its own “decisions” about what it produces. 

And those decisions get to the very core of what it means to create something. I know that many will disagree, but the US Copyright Office, in a January 2025 report, laid it out succinctly. Fully AI-generated works do not qualify for copyright protection because only humans can hold a copyright and works generated by AI lack the human creativity to be considered human-created. 

To be clear, plagiarism and copyright are two very different (but related) things. However, the question of authorship is a space where they overlap. It cuts to the very core of what it means to be the creator of something, which sits at the heart of both plagiarism and copyright issues.

Things do get much more complicated when you look at the gradient of AI usage. The Copyright Office acknowledges that these issues will have to be addressed on a case-by-case basis. However, it still requires those using AI to disclose what elements are AI-generated and exclude them from any application. 

Still, there is a big gray area here, and we are not going to have good answers for a long time, if ever. 

But that does not change a simple fact: If you let AI make creative or intellectual decisions on your behalf and do not disclose it, you are presenting the creative or intellectual work of another as your own. It is that simple. 

To repeat, the issue is not that AI cannot hold a copyright; the issue is that AI-generated works cannot be copyright protected. That is because AI is doing the creative and intellectual labor, not the human who prompted it. No human creativity, no copyright. 

Exceptions to the Rule 

That said, there are going to be exceptions to this rule. Since plagiarism is all about audience expectation, any place where the audience expects AI work is a different matter. 

For example, in an AI art contest, it is the opposite. The audience and judges are expecting AI art so now human-created art becomes problematic. As humorous as that is, it speaks about the same issue.  

There are and always will be spaces where AI creations are accepted and welcomed. Those spaces are allowed to create their own standards with their own audience expectations.  

The problem is when AI work leaves those spaces and is shared outside of that context. What may have started as a perfectly acceptable and transparent posting of an AI work now becomes deceptive. Often, through no fault of the original uploader. 

The other possibility is that audience expectations will change. Citation standards are fluid and change over time. It is possible that, years in the future, the default expectation will be that works are AI-generated, and it will be human-created works that require a specific callout. 

But that is not where we are today. As such, if you put an AI-generated work in a space where others expect human-created content, you are (directly or indirectly) taking credit for creativity and work that you did not do. You are exploiting the audience’s expectations to get more credit than you deserve. 

It is that simple. 

Bottom Line 

AI disclosure and/or AI watermarking should not be controversial. If AI advocates believe that AI-generated work is every bit as valid as human-created ones, then transparency would let the world see that. 

However, there is a broad understanding that, in the court of public opinion, that is not true. The pushback against AI is very strong right now. There is a myriad of reasons for that, but most people clearly consider AI work to be inferior, justly, or unjustly. 

But the solution to that is not to try and pass off AI work as human-created. The answer is to build AI systems and find use cases that benefit human lives. 

If AI-generated content cannot survive transparency, then it does not deserve to exist. Those who publish it without that transparency deserve to be accused of plagiarism. They are presenting a piece of work as their creativity and intellect when it is not.  

In the end, that is my biggest issue with AI. Not that it exists or that it helps people complete tasks more quickly, but that it is so often presented as something it is not. 

AI’s deception may be just one of its sins, but as someone who is passionate about plagiarism, it is the one that gnaws at me the most.