This paper introduces the Sovereign Context Protocol (SCP), an open-source protocol for attribution-aware data access in large language model (LLM) systems. Current approaches to data attribution operate either at the model-internals level or through legal and policy frameworks, but lack a runtime mechanism for tracking how human-generated content is accessed and used. SCP defines a standardized interface between LLMs and creator-owned data, enabling logging, licensing, and attribution of every access event. The protocol specifies six core methods, supports REST and MCP-compatible interfaces, and includes mechanisms for trust scoring, authenticity verification, and access auditing. We present a reference architecture, threat model, and preliminary performance benchmarks, and position SCP within emerging regulatory frameworks such as the EU AI Act. This work argues for a protocol-level approach to attribution, making provenance and accountability a default property of data access in LLM pipelines.
Panchigar et al. (Fri,) studied this question.