---
id: "concept-personal-ai-agents"
type: "concept"
source_timestamps: ["¶4", "§ What Could Go Wrong?"]
tags: ["consumer-ai", "automation", "delegation"]
related: ["concept-agentic-ai", "concept-ai-fiduciary-duty", "concept-retail-manipulation-ai", "prereq-principal-agent-problem"]
definition: "Autonomous AI systems that act as self-directed digital assistants for individuals, capable of executing complex tasks, negotiating, and committing financial resources on the user's behalf."
sources: ["governance"]
sourceVaultSlug: "hbr-seg-governance"
originDay: 7
articleStem: "hbr-cl-88-can-ai-agents-be-trusted"
sourceUrl: "https://hbr.org/2025/05/can-ai-agents-be-trusted"
sourceTitle: "Can AI Agents Be Trusted?"
---
# Personal AI Agents

Personal AI agents are a specific application of [[concept-agentic-ai-d7]] designed to take self-directed action on behalf of an individual user. They function as digital personal assistants capable of handling complex, multi-step workflows such as calendar management, directed research, content curation, and basic communications. Crucially, they possess the ability to find, negotiate for, and purchase goods and services, effectively committing the user's financial resources.

Because they act as a proxy for the user in the real world, they introduce severe principal-agent risks (see [[prereq-principal-agent-problem]]), necessitating—the authors argue—the same level of background checking, insurance, and legal obligation required when hiring a human employee or contractor. The concrete failure modes are catalogued elsewhere in this vault: [[concept-retail-manipulation-ai]], [[concept-sponsor-preference-ai]], and [[claim-ai-vulnerable-to-hacking]]. The proposed remedy is to bind them via [[concept-ai-fiduciary-duty]] within the broader [[framework-trustworthy-ai-triad]].