This project addresses the digital divide by focusing on text generation for people with cognitive disabilities. It introduces the ETR-fr dataset, which contains 523 pairs of French texts aligned with the Easy-to-Read (ETR) European standard. ETR texts use simple vocabulary, clear structure, concrete examples, and contextual summaries to meet the needs of this population. Additionally, the project presents a generative model fine-tuned using LoRA on the mBARTHez backbone. The model demonstrates strong performance in generating ETR content compared to full fine-tuning and Prefix-Tuning on BARTHez, Mistral-7B and mBART. The findings highlight the effectiveness of computer-aided text generation for this purpose.