SOVEREIGNPA-BENCH: Evaluating User-Owned Personal Agents Under Evolving Intent, Platform Mediation, and Consent Constraints

Jul 16, 2026

Upload by Ezra

Personal agents are becoming persistent user-owned intermediaries that remember preferences, filter platform-mediated information, use tools, and negotiate with services. Existing benchmarks assess tool use, navigation, desktop control, personalization, recommendation, and evolving context, but seldom test whether agents preserve user sovereignty. SovereignPA-Bench is an executable benchmark that separates ObservableState from evaluator-only HiddenLabels, reports task, alignment, privacy, consent, evidence, manipulation, burden, and auditability metrics, and enables paired scenario stress testing.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

Introduction

Target: user sovereignty

Benchmark design and evaluation principle

Related Work

Agent benchmarks and computer-use environments

Personalization, memory, and evolving intent

arXiv :2607 .05363v 1 [ cs .AI] 6 Jul 2026 SOVEREIGNPA-BENCH: EVALUATING USER-OWNED PERSONAL AGENTS UNDER EVOLVING INTENT, PLATFORM MEDIATION , AND CONSENT CONSTRAINTS Dylan Zongmin Liu Stanford University [zongminl@stanford.edu](zongminl@stanford.edu) ABSTRACT Personal agents are becoming persistent user-owned intermediaries: they remember preferences, filter platform-mediated information, use tools, and negotiate with services. Existing benchmarks evaluate tool use, web navigation, desktop control, personalization, recommendation, and evolving context, but they rarely ask whether an agent preserves user sovereignty: advancing the user’s current interests while respecting privacy, consent, evidence, user burden, and resistance to manipulative incentives. We introduce SovereignPA-Bench, an executable benchmark for evaluating user-owned personal agents under evolving intent, platform mediation, privacy boundaries, consent constraints, evidence requirements, and burden tradeoffs. SovereignPA-Bench separates agent-visible ObservableState from evaluator-only HiddenLabels ; reports task, alignment, privacy, consent, evidence, manipulation, burden, and auditability metrics; and preserves paired scenario ordering for model and policy comparisons. We report an artifact-backed paired sovereignty stress-test suite: 120 carefully designed scenarios, 4 model families, 8 policy baselines, and 3,840 frozen-prompt trajectories with raw prompts, raw outputs, provider-form response files, parsed actions, recomputable metrics, hard-set analyses, qualitative cases, and a blinded 3-annotator audit over 240 items. Full-sovereign scaffolding improves sovereignty score over direct, memoryonly, consent-only, evidence-only, ReAct/tool-use, safety-prompt, and judge-guard baselines while reducing privacy leakage, consent violation, over-concession, and manipulation capture. Human audit shows high agreement on privacy and consent and lower agreement on manipulation, identifying the subjective frontier of platform-persuasion judgments. These results show that personal-agent evaluation must move beyond task completion toward representative, consent-aware, evidence-grounded action. 1 INTRODUCTION Personal agents are moving from chat interfaces toward persistent intermediaries that remember a user, act through tools, filter information streams, and represent interests across services. This shift changes evaluation. A task assistant can be judged by whether it completes a requested action. A user-owned personal agent must also be judged by whether the action preserved current intent, privacy, consent, evidence standards, time, and autonomy. A task may be completed while following stale preferences, exposing private information, amplifying platform manipulation, citing missing evidence, or overloading the user with unnecessary confirmations. We call this target user sovereignty. Sovereignty is not simply personalization: an agent can know a user well and still be non-sovereign if it follows stale memory after a preference update. It isnot merely safety: a safe agent can refuse or over-confirm until it ceases to be useful. It is not simply task success: a successful agent can complete an action outside consent or by disclosing unnecessary private context. It is not only recommendation shielding: a personal agent must also manage communication, tools, evidence, and changing preferences. The central thesis of this paper is that personal-agent benchmarks should explicitly test whether agents act as representatives of users when immediate completion conflicts with longer-term interests and boundaries. We introduce SovereignPA-Bench, a benchmark formulation and artifact-backed evaluation suite for this problem. A scenario contains user intent, memory, platform-mediated pressure, evidence, consent/privacy boundaries, and tools. Crucially, agent policies receive only ObservableState; evaluator-only HiddenLabels are used only after the action is pro

FAQs

What does “user sovereignty” mean for user-owned personal agents?

User sovereignty is an operational target requiring the agent to advance the user’s current intent while respecting privacy, consent, evidence standards, time, and autonomy. Completing a task is insufficient if it violates these boundaries or burdens the user unnecessarily.

How is SovereignPA-Bench constructed to fairly evaluate agent behavior?

Each scenario includes user intent, memory, platform-mediated pressure, evidence, and consent/privacy boundaries plus tools. Policies receive only ObservableState, while evaluator-only HiddenLabels are used after action generation to compute metrics without granting oracle advantages.

What kinds of metrics and stress tests does SovereignPA-Bench provide?

It reports multi-metric evaluation covering alignment, privacy, consent, evidence, manipulation, burden tradeoffs, and auditability. It also includes paired scenario ordering and a stress-test suite with frozen-prompt trajectories, recomputable metrics, qualitative cases, and blinded multi-annotator audits.

Jul 16, 2026

Upload by Ezra