ChatGPT, developed by the software company OpenAI, has taken the world by storm in recent months.

The online software tool has succeeded over previous attempts at creating a chatbot[1] with passable conversational abilities. Much has been written about it across all forms of media, with views both positive and negative being expressed about ChatGPT, applications like it, such as cactus.ai and perplexity.ai and potential impact of recent NLP developments on society.

ChatGPT is a NLP (Natural Language Processing) model which interacts in a ‘conversational way’.

So first, in case you’ve missed it, ChatGPT is one variety of a functional neurolinguistic programming (NLP) model, and the software program is capable of mimicking the conversational ability of a human. This conversational capability, combined with the efficacy of its responses, has allowed ChatGPT to break through into mainstream appeal beyond the novel curiosity of tech enthusiasts.

Its ability to hold a conversation and respond on almost any subject that is public information[2] sets it apart. Is capabilities range from conducting simple conversations to performing translations, generating essays, simplifying explanations in fields as broad as science to pop-culture, to writing software code or to changing the tone of expression of a submitted piece of prose [3].

It is the opinion of this author as well as many others, that this appeal is lasting, and not a flash in a pan like many other prior attempts such as with the various iterations of Deep Mind, and thus may represent a significant paradigm shift with how we view AI language models[4]. Indeed, there has a been a recent call[5] to halt any further development(s) until the future impact, expressed above, has been quantified and understood.

As with all these things, whether ChatGPT sticks around or not, the technology is here and more likely than not, will be developed further. As such, there are positives, negatives, and several risks, including IP-related issues as a business owner that you should be aware of.

 

The Good…

As mentioned previously, one of the main uses of ChatGPT is to generate content from a prompt, eg. Write me a letter of condolence for a colleague whose mother has just passed away; Tell me a story about a dragon that has no connection to Tolkien; Prepare a CV and resume for me to apply for a job as a business administrator.

ChatGPT can also answer questions around historical events [6], providing facts and situational context, tell jokes, as well as explain complex topics in simplistic language, e.g., Explain wormholes like I’m 5. It can even understand one of the pinnacles of human communication, ASCII art:

 

 

 

The Bad…

However, this quickly reveals the first of the primary pitfalls in using ChatGPT and how it can affect your business. When ChatGPT is wrong about something, it’s very confidently and articulately wrong about it.

 

 

That is clearly a badly drawn snake.

 

The (IP) Ugly…

As one can imagine, in a business context these fallibilities pose risks and some of them have already hit the headlines, such as the misconduct allegations faced by a US gymnast, over her use of cactus.ai [7]. Businesses owners therefore need to be aware of the risks of using content generated tools such as ChatGPT as well as cactus.ai.

The first is a recognition that while it may improve productivity and bring a whole host of improvements to a workforce, it is neither omniscient nor infallible. In fact, it makes trivial errors all the time and when it makes errors it’s often quite good at making them believable – as demonstrated above and in many examples posted online.

Many of these errors can be directly attributed to the data it has been trained on. Any work it produces may require peer review for quality assurance, and that peer review process may take as long as if the reviewer produced the work themselves in the first place.

The second risk relates to how to manage intellectual property concerns, mainly around copyright.

Literary works and software code subsists as copyright through an extensive series of international treaties and national laws. A work must be creative, original and have an author. Currently, AI is not a legal entity so it cannot presently be recognised as an author of a work in Europe or the United States, so cannot own intangible assets.

However, this creates an issue as ChatGPT is able to generate elaborate answers of text from its database of information that it has been trained on and appears to be capable of generating a answer where almost every word in a sentence is potentially different, which may arguably result in a different expression of the work. This interpretation is a legally grey area, as what is considered “work” under copyright is typically an expression of the creative choices by an author.

The company OpenAI that owns ChatGPT and employs the software engineers that developed it currently stakes a claim of ownership through their terms and conditions of use, though they then try to assign that content over to a user, as defined in their Terms of Use:

 Input and Output are collectively “Content.” As between the parties and to the extent permitted by applicable law, you own all Input. Subject to your compliance with these Terms, OpenAI hereby assigns to you all its right, title and interest in and to Output. This means you can use Content for any purpose, including commercial purposes such as sale or publication, if you comply with these Terms.

Source: Terms of Use, OpenAI

Alternatively, perhaps it’s the vast number of original authors of work that ChatGPT’s answer is at least partially derived from as it trains its responses.  Perhaps there is no author, that the work merely exists free of IP constraints: as the case of infinite monkeys locked in a room randomly hitting keys on a typewriter for an infinite amount of time will eventually arrive at all human literary works without any creative expression.

Whatever legal answers eventually emerge to these questions, at the moment businesses have to risk assess based on the uncertainty in this space. If one of your employees, such as a software engineer uses code taken from ChatGPT, there is a very significant risk your business does not own the copyright in it. This means that if your business inserts this code into a product for sale or publication, without the appropriate licenses, your business may be open to prosecution for violation of a third party’s copyright.

Additionally, it may be that ChatGPT is directly plagiarising content. If the legal interpretation is that the content created by ChatGPT is in fact a copy of another’s work, then a user may be committing an act of infringement of an other’s copyright.

Indeed, it is important to note that content created by ChatGPT is derived from content that has been previously generated by others and therefore it is not clear what are the implications in terms of copyright for reusing this content: when is the output “inspired” from existing works and when is it actually infringing them? Where to draw the limit? A ChatGPT user may therefore unwillingly end up infringing someone’s copyright if it publishes output that resembles an existing work too closely. On this point, OpenAI themselves state[8] that:

“it’s up to users to ensure the way they use that content does not violate any laws. The terms and conditions are also subject to change, so do not carry the stability and force of a legal right such as copyright.”

The authors of this article are not sure how this is remotely possible, given users have no real way of back tracing the origin of the program’s responses.

 

Thirdly, there’s the privacy and disclosure issue, as well documented in this recent incident [9].

Samsung employees allegedly leak proprietary information via ChatGPT

Source: https://www.cshub.com/

For those users who input company or proprietary information into ChatGPT, they risk leaking sensitive information to a third party. If you are developing potentially patentable subject matter there are novelty and inventive step requirements that must be met for a patent to be granted. Novelty is assessed against the prior art (all publicly available material) which can include the inventor’s own disclosures. By placing information into ChatGPT the business risks creating a novelty-destroying disclosure that would have an impact on the validity of patent protection.

 

Conclusion…

The allure of enhanced productivity from ChatGPT or similar products may ultimately entice businesses to utilise its capabilities; but those businesses that do so should remain vigilant to the risks posed to their quality control and to the legal intellectual property minefield they are stepping into.

For further information on the IP risks relating to ChatGPT, copyright and other IP issues, please contact the authors via the following links.

 

Suzanne Oliver, Euan Devlin, Gemma Clark

 

Sources:

[1] https://en.wikipedia.org/wiki/Tay_(chatbot)

[2] For various version of ChatGPT there are gaps in its knowledge and its databases contain data only up to 2021. Not all NLP tools or future versions will have this limitation however.

[3] The default writing structure of ChatGPT, when delivering information, is bullet points. However, you can ask it to change the format or tone it uses by amending your statement or question to include this request. You can also request for it to use a formal or informal tone.

[4] For those tech enthusiasts wanting to know how ChatGPT performs this function, and some insight into its inner workings this concise summary by Stephen Wolfram is great place to start.

[5] https://www.cbsnews.com/sanfrancisco/news/musk-scientists-call-for-halt-to-ai-race-sparked-by-chatgpt/

[6] Though the accuracy of the response may be limited to the 2021 cut off – depending which platform you are using and which version of course,

[7} LSU star gymnast used Tik Tok to promote the tool in a paid promotional post, which the university warned could amount to academic misconduct, as it could be considered cheating.

[8] Users are responsible for the way their content is used, not Open AI

[9] Three separate employees are alleged to have leaked information via ChatGPT

 

 

 

At Scintilla, we help innovative companies get a grip on their intellectual property. Our unique commercial approach combines registration of patents and trade marks with strategic input so that IP can be a springboard for business growth. If you would like to discuss your IP needs, do contact us or book a free initial consultation!