The field of artificial intelligence is moving towards a more privacy-conscious approach, with recent developments focusing on protecting sensitive information in language models and surgical modeling. Researchers are exploring innovative methods to balance personalization with privacy risk, including the use of discrete diffusion models and stochastic transformations to anonymize data. The concept of intrinsic dimension is also being investigated as a geometric proxy for the structural complexity of sequences in latent space, which can help mitigate memorization and unintended data leakage. Furthermore, novel approaches such as Private Memorization Editing are being proposed to turn memorization into a defense mechanism to strengthen data privacy in large language models. Noteworthy papers include: Surgeon Style Fingerprinting and Privacy Risk Quantification via Discrete Diffusion Models, which proposes a novel approach to model fine-grained surgeon-specific fingerprinting in robotic surgery. Learning Obfuscations Of LLM Embedding Sequences: Stained Glass Transform, which introduces a learned transformation to provide privacy to the input of large language models. Private Memorization Editing, which turns memorization into a defense to strengthen data privacy in large language models.