In previous articles, I’ve covered much of the good and evil around artificial intelligence (AI). An area that I haven’t covered is the impact of copyright on the use of AI.
UK and EU tackle AI copyright
In the UK, lawmakers are actively discussing how to apply copyright when AI companies train their models using data scraped from the internet. After a lengthy debate over the use of copyrighted data, both the House of Commons and the House of Lords finally passed the Data (Use and Access) Bill. However, the bill ultimately lacked a critical clause that would have protected the rights of original owners.
In contrast, the EU’s Artificial Intelligence Act takes a firmer stance, stating that providers of general-purpose AI (GPAI) models must respect the rights of text and data mining held by relevant rightsholders. These providers must obtain the necessary authorizations if they wish to mine the text and data of these rightsholders. However, this requirement hinges on the original copyright holders clearly expressing their views on the usage of any material they have posted in the public domain.
US copyright cases put AI training under pressure
Meanwhile, in the US, courts are currently hearing several AI-related copyright cases. Although these cases initially appeared to have some potential for success, it seems increasingly unlikely that they will prevail as time goes on. The primary reason for this shift is a pre-publication version of a key document from the US Copyright Office, which specifically examined copyright issues related to AI training. Published in May 2025, this document appeared to favor original creators in many instances, making it more challenging for AI engine vendors to use original content in their training models.
However, the Trump Administration dismissed the Librarian of Congress just before the document’s publication and subsequently fired the Register of Copyrights shortly after. At the same time, AI vendors are vigorously lobbying for an open market where they consider anything published in the public domain to be fair game. It is unfortunately not difficult to predict the direction the Trump Administration may take.
If the US decides that anything and everything on the web, including content behind paywalls or secure walls, counts as fair game for AI training, many other countries are likely to follow suit. The prevailing belief is that AI represents the new battleground, and a country can only thrive if it operates under the same rulebook as the lowest common denominator. The impact on original creators could be significant, leading many to question the value of their creative efforts. Why invest weeks or months in producing a book, song, video, or image if a generative AI system can produce something “good enough” in seconds?
Why registration matters in copyright enforcement
Many creators have a sufficient understanding of the copyright market, which could complicate the use of AI-generated creatives. One example, although not directly related to AI, involves an American photo artist who captured images showing the number of planes taking off at various airports across the US. The artist knows he holds the original copyright to his work, but if someone else were to use it and the case went to court, the courts would first assess whether the use was accidental or unintentional, and only then would they consider any financial loss to the copyright holder.
If the court rules a financial loss, it will determine the compensation owed to the artist. This could be based on the fee he would typically charge a customer for using a digital asset, usually in the low hundreds of dollars, or it could be calculated based on a market rate, generally within the same range. It is clear that the costs of pursuing such a case would far outweigh the potential benefits.
The US Copyright Office provides an additional level of copyright protection, it allows for creative assets to be fully registered, providing a significantly higher level of protection. Any violation of a registered copyright can lead to punitive damages, which may include fines in the thousands of dollars and, in some cases, jail time.
AI copyright risks for MSPs
So, how does this impact a managed service provider (MSP)? AI engine providers are likely to obfuscate and contest any copyright claims before filing a lawsuit, prolonging the court process as much as possible. But what about you as an MSP?
Let’s say you use a generative AI engine to create some text, and someone points out that it closely resembles their registered copyright asset. What if the same happens with an AI-generated picture or video? It is highly unlikely that the registered copyright holder will pursue lawsuits against large companies like Microsoft, Google, OpenAI, or Grok; instead, they will likely target individuals. After all, it was your organization that provided the prompts and details for creating the AI asset. Even if you were unaware of how that asset was generated, you should still be mindful of the process.
The key takeaway from this discussion is for MSPs (and everyone) to exercise caution when using AI to create digital assets. Thoroughly check reliable search engines to see if similar content already exists. If you find something, reach out to the original copyright holder to request permission, which is often available at minimal cost, with proper acknowledgment.
Photo: aijiro / Shutterstock
This post originally appeared on Smarter MSP.