Currently, works created by artificial intelligence (AI) such as ChatGPT — even if a human wrote a text prompt to get the output — isn’t technically protected by copyright. It’s really nuanced, and we’re watching closely to see if anything changes.
The problem is, when it comes to training Large Language Model (LLMs) and AI it’s okay for them to use things that have copyright to train the systems. There’s a law saying it’s fair to use copyrighted stuff in certain ways without asking the owner. However, there are some arguments in courts that might change this.
Over the last two years, Generative AI has quickly changed how we live and create. But this has brought up so many tricky legal questions about who owns what and what’s fair. It’s making us rethink what it means to own something and how creativity works.
The ChatGPT copyright dilemma
People are questioning if ChatGPT should be allowed to use stuff created by others to form its responses. Another question is whether only humans can be considered the creators of AI-generated content, or if the AI itself can be seen as the creator, especially when it’s being creative.
Let’s tackle the first question. The tech behind ChatGPT is known as a Large Language Model (LLM). To get better, it reads lots of data, like info from websites and books. In the UK, AI developers can use this for non-commercial reasons. OpenAI, the company behind ChatGPT, says users own what it creates, but they need to make sure they’re not breaking any laws. The rules might also change, so they don’t have the same stability as copyright.
The only real solution is to make clear laws and rules. Otherwise, every group will have to go to court separately, trying to prove they own what the AI uses. If governments don’t act, we might end up in a situation where everyone uses copyrighted stuff without the original creator saying it’s okay.
Now, onto question two: who can own what the AI creates. If the original owner doesn’t claim rights over what the AI used, it might be that regular users or the companies behind the AI can say it’s theirs.
But here’s the thing about copyright laws: they usually say only stuff made by humans can be protected. The codes behind ChatGPT were made by people at OpenAI, so the company might have rights over that. But when it comes to what the chatbot says, it gets tricky.
There’s another idea: the AI itself owning the stuff it makes. But in the UK, the law says AIs can’t own copyright because they’re not human. It’s unlikely this will change soon, given the government’s stance on AI.
Right now, policymakers are sticking to humans as the ones who get copyright. But as AI gets smarter, they might think about giving legal rights to AIs. This would be a big change in how copyright works and who (or what) can be called the owner of copyright.
If this happens, businesses using AI in their products and services, like Microsoft with its Copilot based on ChatGPT, will need to deal with these changes.
Who owns the copyright in AI-generated work?
The big question right now is making the rounds, and as the UK Government thinks it over, the answer isn’t clear. According to the Copyright Designs & Patents Act 1988, for computer-generated artwork, the ‘author’ (usually the first copyright owner) is considered to be “the person who arranged for the creation of the work.”
In the UK, based on the Copyright, Designs and Patents Act 1988, art made by a computer can have copyright protection. There’s no need to register this copyright in the UK. The law says a “computer-generated” work is something created by a computer when there’s no human creator. The “author” of a computer-generated work is the one who made the plans to create it. These works are protected for 50 years from when they were made. So, in the UK, the idea of who the creator is and who arranged for it to be made are separate things.
Who owns the rights to AI-generated work is a tricky question:
- Is it the person using the AI art generator, providing the inputs?
- The one who made the software?
- The person who taught the AI-powered software?
- Or maybe the people who own the copyright for the elements used to make the new artwork?
There’s no clear answer yet. It’s still being debated, and until it’s figured out (hopefully in the next year or two), any legal fights over this will need careful examination case by case.
When the UK Government asked for opinions on this, many thought the creator of the AI-generator software should own the copyright for what it produces.
But here’s the twist: training the AI model behind the software needs a ton of data, and that data is usually not public (it’s often a trade secret). So, who’s to say that true copyright ownership isn’t shared among many, including the creators of the original work used to train the AI model and, in the end, make the new content?
Do brands have legal ownership over AI-Generated content?
Brands using AI for content need to deal with legal questions about who owns what. If you spend time and money on making or teaching an AI to create stuff, you might have a right to the copyright. But things can get confusing legally, which is why having clear agreements and good records becomes even more crucial.
Can AI infringe copyright?
The AI itself can’t get in trouble, but the people using it might face copyright problems.
The responsibility for copyright infringement often falls on the human entities that create, deploy, or use AI systems rather than on the AI itself. If an AI system generates content that infringes copyright, the legal accountability typically rests with the individuals or organisations involved in developing, training, and implementing the AI.
In some jurisdictions, the legal framework might consider the actions of the human actors behind the AI when determining copyright infringement. As AI technologies continue to advance, legal systems may evolve to address these complex issues.
Legal Actions Involving ChatGPT and OpenAI:
- A significant legal showdown is underway in a New York federal court, shaping the future of ChatGPT and comparable AI products that heavily rely on copyrighted human content. The central issue is whether these AI chatbots, specifically those created by OpenAI in collaboration with Microsoft, are in violation of copyright and fair competition laws.
- Well-known authors have initiated at least three lawsuits against OpenAI, claiming the unauthorized use of their copyrighted work in training the chatbot, which generates text based on user inputs.
What ChatGPT have said in response?
Instead of removing copyrighted content from ChatGPT’s training data, the creator of the chatbot is proposing to cover legal expenses for clients facing copyright infringement lawsuits.
OpenAI CEO Sam Altman stated on Monday, “We can defend our customers and pay the costs incurred if you face legal claims around copyright infringement, and this applies both to ChatGPT Enterprise and the API.” This compensation initiative, named Copyright Shield, is applicable to users of the business tier, ChatGPT Enterprise, and developers utilising ChatGPT’s application programming interface. Users of the free versions, ChatGPT and ChatGPT+, are not included.
Current position on copyright in the UK for ChatGPT:
The recent interim report from the Science, Innovation, and Technology Select Committee sheds light on the intellectual property and copyright challenges brought about by AI. These challenges underscore the importance of safeguarding the rights of content creators as AI utilizes copyrighted materials, especially through activities like web scraping and generating new content.
A suggested solution involves collaboration between the creative industry and the AI sector to establish a licensing framework for copyrighted materials used in AI training. The UK Intellectual Property Office (UKIPO) is also exploring the possibility of a voluntary code of practice concerning copyright and AI, but details about the progress are not yet available.
Additionally, the Culture, Media, and Sport Select Committee recently released a report on Connected Tech: AI and Creative Technology. The committee advises against adopting proposals from the UKIPO for a broad copyright exemption for commercial text and data mining (TDM). While the UKIPO’s proposals aimed to introduce exceptions to copyright and database rights for commercial TDM, with no opt-out option for rights holders, the committee suggests not moving forward with these suggestions.
Following the UKIPO consultation, the UK government continues to review the law, considering feedback from the oral evidence session. There is a possibility of amending, replacing, or removing protection in the future. As AI advances and becomes more widespread, it will be intriguing to observe how the legal landscape evolves in both the UK and the USA.
JMR Solicitors – Contact us for a free consultation about AI generated content and copyright
If you have any concerns about copyright and how ChatGPT might be infringing your rights, we’re here to help. We’re closely monitoring the situation, and should you be able to put forward a case, we can help you with this. Contact us on 0161 491 3933 or email info@jmrsolicitors.co.uk