Dynaseal: A Backend-Controlled LLM API Key Distribution Scheme with Constrained Invocation Parameters

TL;DR

Dynaseal introduces a backend-controlled API key distribution scheme for LLMs that allows fine-grained constraints on model invocation parameters, improving flexibility and control in edge-device deployments.

Contribution

The paper presents Dynaseal, a novel approach for backend-controlled API key distribution that enables constrained invocation parameters for LLMs.

Findings

Enables fine-grained control over LLM API usage.

Improves flexibility for edge-device deployment.

Enhances backend management of model invocations.

Abstract

Due to the exceptional performance of Large Language Models (LLMs) in diverse downstream tasks,there has been an exponential growth in edge-device requests to cloud-based models.However, the current authentication mechanism using static Bearer Tokens in request headersfails to provide the flexibility and backend control required for edge-device deployments.To address these limitations, we propose Dynaseal,a novel methodology that enables fine-grained backend constraints on model invocations.