Easy-read and large language models: on the ethical dimensions of LLM-based text simplification | Synapse