To develop a source-aware methodology for AI-assisted nursing documentation that preserves who supplied a clinically meaningful proposition, when the underlying event or statement occurred, whether the proposition represents observation or inference, what uncertainty remains, and who is responsible for correction. Generative artificial intelligence can reduce documentation workload and produce fluent clinical summaries, yet linguistic fluency can conceal changes in epistemic status. This problem is especially important in palliative care, where a note may combine patient self-report, relatives’ accounts, direct nursing observations, information inherited from earlier records, and algorithmically inferred statements across a rapidly changing clinical trajectory. Original non-empirical methodological framework developed through critical synthesis of contemporary nursing, palliative-care, clinical-documentation, natural-language-processing, generative-AI, temporal-reasoning, human-factors, and governance scholarship. A critically selected 2017–2026 evidence corpus was used to examine clinical text provenance, serious-illness documentation, temporal information, AI-generated summaries, uncertainty, human editing, nursing workflow, and accountability. The evidence set was used as a conceptual foundation and was not treated as a systematic-review sample. The central methodological unit is the clinically meaningful proposition within a nursing note. Each proposition should retain an identifiable source class, temporal anchor, epistemic status, uncertainty state, and correction pathway. Patient self-report, relatives’ interpretation, clinician observation, imported historical information, and algorithmic inference therefore remain distinguishable even when they are compressed into a single generated note. Trustworthy AI-assisted palliative-care documentation depends on preserving the provenance and current interpretability of clinical statements. A generated sentence should not acquire greater authority merely because an AI system has rendered it concise, coherent, or definitive.