"Human-AI interaction and the development of an Online Safety Code for Mental Health"

The development of artificial intelligence tools presents both positives and negatives for wider society. However, in the mental health space there are concerns that applications such as ChatGPT and others facilitate self-harm especially among young adults. To address these concerns, this paper examines the possibility of utilizing ChatGPT as a safeguard against self-harm rather than as a facilitator of self-harm.

To achieve this objective the current context limits of ChatGPT are tested when discussing mental health concerns. The next step for this research then includes developing a code of conduct based on Maslow's Hierarchy of Needs with ChatGPT collaboration. This code seeks to strike the right balance between improving the use of ChatGPT as a mental health support but also the need for this application to divert a user to real-world supports if necessary.

The final objective of this research is to develop a tiered approach to how an online AI application such as ChatGPT can successfully aid those with mental health problems within certain safeguards. These safeguards are based on Maslow's Hierarchy of Needs and this should ensure that once a certain threshold has been reached that real-world support mechanisms are then promoted by the online application.

Resource constraints across health care sectors in Western Europe and further afield, point to the need for a nuance approach when utilizing AI for mental health supports. While online applications can not replace the role of mental health professionals these models may under the right controls resolve resource constraints in this sector.