Title: Using LLM-based ‘chatbots’ to support unpaid carers of people living with dementia at home: A human rights perspective Abstract: Large Language Models (LLM) like OpenAI's ChatGPT are transforming the way people work, including in health and social care professions. They also have the potential to support family carers with everyday care tasks, for example by providing quick solutions to care related challenges. This could have an important impact on carers' sense of autonomy by providing them with round the clock virtual assistance in their role. Various organisations with the aim to support carers, for example of people living with dementia, are currently working on integrating LLM’s into their service offer. This is in addition to content available on a website and person-to-person support through phone helplines. But such use of LLMs in social care poses some fundamental questions and challenges regarding the reliability and quality of responses and solutions. There is a real need to understand the ethical, legal, and practical ramifications of applying LLMs to provide support to family carers, especially when it comes to capacity and healthcare related questions. This presentation reports on a small study testing out responses generated by ChatGPT to some fundamental care tasks, including on mental capacity, and comparing them with advice generated by care professionals. It suggests that generated responses, although filling a gap in service round the clock service provision, require quality control from trained professionals and considerations for regulation that expand into digital social care provision.