Search papers, labs, and topics across Lattice.
This paper proposes a Trusted Execution Environment (TEE)-based architecture for AI-assisted grant evaluation to ensure auditability without exposing the model or scoring rubric. The architecture uses remote attestation to allow external verification of the model, rubric, prompt, and input representation used in the evaluation process. The system generates an attested evaluation bundle, a signed record linking submission hashes, input hashes, model measurements, and evaluation outputs, while also incorporating a sanitization layer to mitigate prompt injection risks.
You can now audit AI-assisted grant evaluations without revealing the model's secrets, thanks to a clever TEE-based architecture that cryptographically proves what happened inside.
Public agencies are beginning to consider large language models (LLMs) as decision-support tools for grant evaluation. This creates a practical governance problem: the model and scoring rubric should not be exposed in a way that allows applicants to optimize against them, yet the evaluation process must remain auditable, contestable, and accountable. We propose a TEE-based architecture that helps reconcile these requirements through remote attestation. The architecture allows an external verifier to check which model, rubric, prompt template, and input representation were used, without exposing model weights, proprietary scoring logic, or intermediate reasoning to applicants or infrastructure operators. The main artifact is an attested evaluation bundle: a signed, timestamped record linking the original submission hash, the canonical input hash, the model-and-rubric measurement, and the evaluation output. The paper also considers a scenario-specific prompt injection risk: applicant-controlled documents may contain hidden or indirect instructions intended to influence the LLM evaluator. We therefore include a canonicalization and sanitization layer that normalizes document representations and records suspicious transformations before inference. We position the design relative to confidential AI inference, attestable AI audits, zero-knowledge machine learning, algorithmic accountability, and AI-assisted peer review. The resulting claim is deliberately narrow: remote attestation does not prove that an evaluation is fair or scientifically correct, but it can make part of the evaluation process externally verifiable.
