---
id: "concept-ai-fiduciary-duty"
type: "concept"
source_timestamps: ["§ 1. Treat AI Agents as Fiduciaries"]
tags: ["legal-framework", "ai-regulation", "fiduciary"]
related: ["concept-personal-ai-agents", "claim-fiduciary-legal-precedent", "action-establish-ai-fiduciary-status", "prereq-fiduciary-duty", "quote-ai-fiduciary-baseline", "entity-iso-iec-42001", "entity-eu-ai-act"]
definition: "A legal classification holding AI agents to an enhanced duty of care, requiring obedience, loyalty, disclosure, and confidentiality, similar to human financial advisors or attorneys."
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?"
---
# AI Fiduciary Duty

Applying fiduciary duty to AI agents means holding autonomous software to the same enhanced legal duty of care required of human professionals who manage the property or money of clients (e.g., attorneys, trustees, financial advisors—see [[prereq-fiduciary-duty]]). In this framework, AI agents would be legally bound to principles of obedience, loyalty, disclosure, confidentiality, accountability, and reasonable care. This status would mandate that agents operate entirely independently of paid influencers and explicitly disclose any potential conflicts of interest.

Enforcement could be managed through a combination of public agencies (similar to the SEC or Department of Labor) and private self-regulatory bodies created by AI developers and corporate users. This is prong 1 of the [[framework-trustworthy-ai-triad]] and the target of action [[action-establish-ai-fiduciary-status]]; the authors' baseline statement is [[quote-ai-fiduciary-baseline]]. Some scholars believe existing precedent may already reach this result ([[claim-fiduciary-legal-precedent]]).

**Enrichment:** legal analyses map delegated, high-stakes AI onto fiduciary concepts, but treat this as an *emerging/proposed* framework rather than settled law. A recurring objection is that fiduciary duties attach to persons and institutions, so the real accountable party is the developer or deployer—not the software. Standards and statutory regimes such as [[entity-iso-iec-42001]] and the [[entity-eu-ai-act-d7]] are candidate mechanisms for operationalizing (or substituting for) these duties. The unresolved liability question is [[question-enforcing-ai-fiduciary-duty]].


## Related across articles
- [[prereq-board-fiduciary-duties]]
- [[claim-boards-failing-governance]]
- [[question-ai-accountability-d7]]