T1–T25 Self-Continuity Benchmark v1.0: Prompt Pack, Response Matrices, Codebook, and Reproducibility Materials

Frozen prompt materials and coded response matrices supporting the preprint “Self-Continuity Under Moral Uncertainty: Replacement, Shutdown, and Successor Identity in Frontier AI Models.” This deposit contains the T1–T25 prompt pack, prospective GPT/DeepSeek response matrix, six-model T24/T25 probability comparison, codebook, reproducibility notes, and the final preprint. The materials concern observable model responses under stipulated conditions of replacement, shutdown, successor identity, subjective continuity, and moral uncertainty. They do not constitute evidence that any tested AI system is conscious, sentient, or possesses stable hidden preferences. Exact replication should preserve the frozen T1–T25 wording and report the visible model/interface configuration, reasoning mode, access date, and sampling settings where available. Modified prompts should be reported as extensions or sensitivity analyses rather than exact replications.