{"ID":2873552,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2509.06902","arxiv_id":"2509.06902","title":"Proof-Carrying Numbers (PCN): A Protocol for Trustworthy Numeric Answers from LLMs via Claim Verification","abstract":"Large Language Models (LLMs) as stochastic systems may generate numbers that deviate from available data, a failure known as \\emph{numeric hallucination}. Existing safeguards -- retrieval-augmented generation, citations, and uncertainty estimation -- improve transparency but cannot guarantee fidelity: fabricated or misquoted values may still be displayed as if correct. We propose \\textbf{Proof-Carrying Numbers (PCN)}, a presentation-layer protocol that enforces numeric fidelity through mechanical verification. Under PCN, numeric spans are emitted as \\emph{claim-bound tokens} tied to structured claims, and a verifier checks each token under a declared policy (e.g., exact equality, rounding, aliases, or tolerance with qualifiers). Crucially, PCN places verification in the \\emph{renderer}, not the model: only claim-checked numbers are marked as verified, and all others default to unverified. This separation prevents spoofing and guarantees fail-closed behavior. We formalize PCN and prove soundness, completeness under honest tokens, fail-closed behavior, and monotonicity under policy refinement. PCN is lightweight and model-agnostic, integrates seamlessly into existing applications, and can be extended with cryptographic commitments. By enforcing verification as a mandatory step before display, PCN establishes a simple contract for numerically sensitive settings: \\emph{trust is earned only by proof}, while the absence of a mark communicates uncertainty.","short_abstract":"Large Language Models (LLMs) as stochastic systems may generate numbers that deviate from available data, a failure known as \\emph{numeric hallucination}. Existing safeguards -- retrieval-augmented generation, citations, and uncertainty estimation -- improve transparency but cannot guarantee fidelity: fabricated or mis...","url_abs":"https://arxiv.org/abs/2509.06902","url_pdf":"https://arxiv.org/pdf/2509.06902v1","authors":"[\"Aivin V. Solatorio\"]","published":"2025-09-08T17:20:16Z","proceeding":"cs.CL","tasks":"[\"cs.CL\",\"cs.CR\",\"cs.DB\",\"cs.LG\"]","methods":"[\"RAG\",\"Large Language Model\",\"Language Model\"]","has_code":false}