1 May 2023 | Reading time: 3 minutes
ChatGPT is an artificial intelligence (AI) chat bot, launched by OpenAI in 2022. It uses natural language processing to have human-like conversations with users who can feed it any prompt or question. Within seconds, the bot will reply with a response on almost any topic, based on its access to a staggeringly vast range of information.
For the bot to have a wide breadth of knowledge, during training, developers fed it huge amounts of academia such as articles and blog posts.
Authors and artists are now becoming increasingly concerned that their works may be reproduced in ChatGPT’s responses without permission or credit. Notably, Getty Images has sued a different AI platform for copyright infringement, claiming the AI was fed millions of Getty images to train it.
Like many areas of technology, AI’s rapid development has meant that the law is now playing catch up. One area which has become topical is copyright liability. Is the bot itself infringing other authors’ copyright, or is the liability on the software engineers feeding the information into the system?
No human intelligence in Chat GPT
Another copyright issue relates to whether copyright exists in what ChatGPT produces. Copyright does not protect ideas, but rather the tangible expression of those ideas (such as in books, artwork and musical works). Under the Copyright Act 1968 (Cth), the infringer is the person who does, or authorises the doing of, the infringement of the owner’s copyright.
Australian law does not expressly provide that authors need to be ‘human’ for copyright to be protected, but the courts have shown a strong preference for human authorship. This was evident in the Federal Court which held that a computer-generated phone book could not be protected by copyright, as it lacked human authorship. This means that if a business or individual uses AI to generate content, such content will not be protected by copyright laws, because the author of the work was a computer (and not a human).
In the US, the author must be human for copyright purposes. In the monkey selfie case, a monkey took a selfie on a photographer’s camera, and the photographer then published a book of photos including the monkey selfie. The court established that the monkey could not sue for copyright infringement, and the US copyright office further stated that they would only register an original work if the author is human.
In contrast, the UK has an exception to the human authorship rule for computer generated work, defining the programmer or user of the AI as the author. By defining a human as the author (rather than the AI), the resulting work may be eligible for copyright protection.
The copyright issues are further complicated by the ChatGPT terms, which state:
‘Subject to your compliance with these Terms, OpenAI hereby assigns to you all its right, title and interest in and to Output.’
This implies there are intellectual property rights, in the first place, which are capable of assignment. However, under Australian and US law, there is no copyright for ChatGPT to assign.
Copyright laws being challenged
Given Chat GPT operates autonomously and makes significant independent decisions, copyright laws are increasingly being challenged. As generative AI platforms evolve, it will be important for businesses to think proactively about how these services are used in the workplace. Whether you use ChatGPT for fun or for business use, the message is clear: chat with caution.