Device-Level Access Management for Generative Artificial Intelligence: A Unified Framework for Unidirectional and Bidirectional Device Authorization in AI Conversations

Generative artificial intelligence (GAI) systems increasingly serve as persistent conversational environments in which users create, exchange, retrieve, and maintain potentially sensitive information across multiple devices. Although contemporary platforms commonly provide account authentication, session management, and device-level security mechanisms, these mechanisms do not necessarily provide users with sufficiently granular control over which individual devices can access particular AI conversations, how such access is authorized, or how permissions are independently managed among multiple conversation participants. This paper proposes Device-Level Access Management for Generative Artificial Intelligence (DLAM-GAI), a unified conceptual and technical framework that introduces device-centric authorization as an explicit layer of conversational access governance. The framework distinguishes account authentication from device authorization and conversation-level permission enforcement, allowing users to inspect registered devices, authorize or deny access, assign differentiated permissions, and revoke previously granted access. Its central contribution is the integration of unidirectional and bidirectional device authorization into a common policy model. Under unidirectional authorization, an individual participant independently controls which devices may access their own conversational resources. Under bidirectional authorization, multiple participants independently govern device access associated with their respective identities and protected conversational resources without acquiring unilateral authority over another participant's device permissions. The framework further introduces conversation-scoped authorization, permission isolation, temporary access, revocation propagation, device trust evaluation, and auditable authorization histories. A formal authorization model, system architecture, threat model, operational procedures, and evaluation methodology are developed to establish the technical foundations of the proposed approach. Rather than presenting unperformed experiments as established findings, this study defines measurable security, usability, consistency, and performance criteria for subsequent implementation and empirical validation. DLAM-GAI aims to extend conventional account-centric security toward participant-controlled, device-specific, and conversation-aware access management for generative AI environments.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-16
DOI
https://doi.org/10.5281/zenodo.24278073
Primary Topic
Access Control and Trust
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Device-Level Access Management for Generative Artificial Intelligence: A Unified Framework for Unidirectional and Bidirectional Device Authorization in AI Conversations

Jincheng Zhang
19 links
Zenodo (CERN European Organization for Nuclear Research)
Access Control and Trust
preprint

Device-Level Access Management for Generative Artificial Intelligence: A Unified Framework for Unidirectional and Bidirectional Device Authorization in AI Conversations

Jincheng Zhang
preprint en

Abstract

Generative artificial intelligence (GAI) systems increasingly serve as persistent conversational environments in which users create, exchange, retrieve, and maintain potentially sensitive information across multiple devices. Although contemporary platforms commonly provide account authentication, session management, and device-level security mechanisms, these mechanisms do not necessarily provide users with sufficiently granular control over which individual devices can access particular AI conversations, how such access is authorized, or how permissions are independently managed among multiple conversation participants. This paper proposes Device-Level Access Management for Generative Artificial Intelligence (DLAM-GAI), a unified conceptual and technical framework that introduces device-centric authorization as an explicit layer of conversational access governance. The framework distinguishes account authentication from device authorization and conversation-level permission enforcement, allowing users to inspect registered devices, authorize or deny access, assign differentiated permissions, and revoke previously granted access. Its central contribution is the integration of unidirectional and bidirectional device authorization into a common policy model. Under unidirectional authorization, an individual participant independently controls which devices may access their own conversational resources. Under bidirectional authorization, multiple participants independently govern device access associated with their respective identities and protected conversational resources without acquiring unilateral authority over another participant's device permissions. The framework further introduces conversation-scoped authorization, permission isolation, temporary access, revocation propagation, device trust evaluation, and auditable authorization histories. A formal authorization model, system architecture, threat model, operational procedures, and evaluation methodology are developed to establish the technical foundations of the proposed approach. Rather than presenting unperformed experiments as established findings, this study defines measurable security, usability, consistency, and performance criteria for subsequent implementation and empirical validation. DLAM-GAI aims to extend conventional account-centric security toward participant-controlled, device-specific, and conversation-aware access management for generative AI environments.

Zenodo (CERN European Organization for Nuclear Research)
Access Control and Trust
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon