Introduction
Recently, the Artificial Intelligence (AI) chat bot, ChatGPT (created by OpenAI) has set the internet abuzz a few days after its release. ChatGPT is an AI-based chatbot that is trained to answer complex questions in a simple human-readable[1] manner. One can think of it as an advanced virtual assistant, such as Siri or Alexa. Some of the features include answering complex questions, writing pieces of literature, creating graphics and animations, and composing music. The responses generated by ChatGPT are based on an AI system that has been trained on both natural language and code.[2]
There is much speculation on how it might impact industries. One of the most apparent ways is in the customer service industry. ChatGPT can provided automated support to customers. It also can function as a virtual assistant to manage schedules or make reservations. ChatGPT has a wide knowledge base, which means that it can also perform tasks that it has not been trained to do. The software further includes safeguards that refuse to do inappropriate requests.[3] Despite the many features, ChatGPT is not human, and can be wrong or inaccurate. Further, the knowledge base only includes information up until 2021, and any new information from 2022 onwards will not be included in the results.
Although ChatGPT can assist industries in a multitude of ways, there are a number of legal issues which users should be aware of, as highlighted below.
Legal Issues with ChatGPT
Licenses
In general, any commercial use of generated content requires a license. It is likely that a license from OpenAI is required if a user would like to use the generated responses for commercial purposes. There are currently no terms of use for ChatGPT.
Copyright
There are concerns that generated responses are based on information protected by copyright. If true, this means that ChatGPT produces outputs that may lead to copyright infringement. When using content generated by ChatGPT, the user must acknowledge that the content is created by AI and not human-generated. If you pass off the generated content as your own work, you may be held liable for copyright infringement.
Plagiarism
Plagiarism is where someone presents another’s work or ideas as if it is their own. Academics and teachers face a new problem, where students can request ChatGPT to author an essay or speech. The result is a very human-sounding piece of work.
Personal Data Protection
Since personal information can be processed in the use of ChatGPT, data privacy and protection laws, such as the European General Data Protection Regulation (“GDPR”) and South Africa’s Protection of Personal Information Act (“POPIA”) must be observed. Some of the issues that may arise in the processing of personal information by the chatbot are as follows:
- The personal information processed by ChatGPT may be used for other purposes, such as artificial intelligence training. This is not in line with the purpose limitation principle.
- Privacy principles allow for the right of deletion or the right to be forgotten. Currently, ChatGPT does not allow for prompts given by the user to be deleted.[4]
- GDPR and POPIA both state that personal information must be accurate. It is not clear whether this requirement is adhered to.
- The principle of minimality requires that the bare minimum information must be used to carry out the purpose of controller/responsible party. However, ChatGPT may not adhere to this requirement, because billions of data points are used.
It is evident that further consideration needs to be given to the use of ChatGPT to ensure that the privacy rights of individuals is respected.
Conclusion
ChatGPT appears to be a brilliant piece of technology and is a great example of how AI can impact many industries. However, there is still numerous potential legal issues and questions that arises concerning its use. In many jurisdictions around the world, laws regulating the use of AI is still in its infancy stages. These laws will have to cater for the inherent risks and dangers, as well as the ethical implications relating to the use of AI. A stronger focus on AI regulation is required in order to appropriately address the legal issues.
Contact us for more good, clear, precise advice.
[1] Human-readable means any encoding of data or information that can be naturally read by humans.
[2] https://beta.openai.com/docs/models/gpt-3.
[3] https://beta.openai.com/docs/usage-policies/use-case-policy.
[4] https://help.openai.com/en/articles/6783457-chatgpt-general-faq.
