Par šo projektu

About SAFE

SAFE (Source-aware Argument Fidelity & Evidence) is an open-source protocol for checking facts and claims the way a careful researcher would — by weighing where the evidence actually comes from. Instead of treating every source as equal, SAFE scores each claim against a clear hierarchy: peer-reviewed research and systematic reviews first, weighted by how recent the findings are; university and higher-education level knowledge, weighted by complexity; broader scientific consensus; and news reporting from outlets that cite their sources transparently.

Built as a cross-platform connector rather than a general web-search tool, SAFE is designed to work alongside other AI plug-ins and connectors, giving them a structured, auditable way to judge evidence quality instead of guessing at it. Every score comes with a full audit trail — which sources were used, how they were tiered, and why the final confidence level was assigned — so nothing about the evaluation is a black box.

SAFE is maintained as a volunteer, open-source project, built and shared freely under the Unlicense. If you rely on it, use it in your own tools, or simply believe that AI systems should be able to explain their reasoning, you can support its development below.