AI is only as good as the data behind it, and in healthcare, that data is often messy, outdated, and biased. As systems create digital versions of patients, known as twins, the risks increase when data deteriorates or is used without informed consent. Understanding how data breaks down over time, how it's mislabeled or misused, and why clean, well-governed data matters is essential to creating safer, smarter tools that actually work for people, not against them.