Harvard Law Review Identifies Fiduciary Blind Spots in Frontier AI Board Structures at OpenAI and Anthropic
What happened
Harvard Law Review published Amoral Drift in AI Corporate Governance in its January 2025 print edition, offering a peer-reviewed legal examination of how frontier AI companies structure board authority, fiduciary duty, and stakeholder accountability. The article scrutinizes the governance architectures of OpenAI and Anthropic, both of which use non-standard corporate forms, including OpenAI's capped-profit LLC structure and Anthropic's public benefit corporation model, to balance commercial growth against safety commitments. The authors argue that these hybrid structures create ambiguity about who directors ultimately owe fiduciary duties to, whether shareholders, the public, or a defined safety mission, and whether existing corporate law is adequate to discipline that ambiguity. The article frames this structural uncertainty as "amoral drift": a tendency for AI companies to gradually deprioritize safety commitments as commercial pressures intensify, without any clear legal mechanism to prevent or detect the slide. The analysis has direct implications for enterprise purchasers who rely on vendor safety pledges, and for compliance teams assessing the governance credibility of major AI platform providers.
Why it matters
- ·Vendor safety commitments from frontier AI companies may lack enforceable legal grounding: if a provider's board structure does not create binding fiduciary obligations to uphold its stated safety mission, compliance teams cannot treat those commitments as reliable contractual or reputational controls.
- ·Enterprises with critical dependencies on OpenAI or Anthropic APIs face concentration risk that is amplified by governance opacity: the absence of clear board accountability mechanisms means material changes to safety posture may occur without the external disclosure that would trigger a vendor risk reassessment.
- ·Regulators in multiple jurisdictions are increasingly scrutinizing AI vendor governance as part of third-party and systemic risk frameworks; this legal analysis signals that existing corporate law may not fill the gap, raising the probability that prescriptive vendor governance disclosure requirements will emerge and affect enterprise procurement and diligence programs.
Governance controls affected
What to do now
- ☐Review existing vendor risk assessments for OpenAI, Anthropic, and other frontier AI providers to determine whether your team has documented the legal enforceability of each vendor's stated safety commitments, and flag gaps for immediate remediation.
- ☐Update your AI vendor contract requirements (PRC-002) to include explicit representations about governance structure, board composition changes, and any amendments to the vendor's safety or mission commitments, with notification obligations triggered by material changes.
- ☐Assess vendor concentration risk for any frontier AI platform used in critical workflows and document the degree to which your organization's risk posture depends on the vendor's self-reported safety governance remaining intact.
- ☐Escalate the Harvard Law Review findings to your board AI risk committee as evidence supporting enhanced vendor governance monitoring and, where relevant, revised AI risk tolerance documentation.
- ☐Brief legal counsel on the fiduciary ambiguity described in the article and determine whether your current vendor agreements provide sufficient contractual recourse if a frontier AI provider materially changes its safety posture following a corporate restructuring.
What to watch next
Compliance teams should monitor whether state attorneys general in Delaware, California, or other jurisdictions where these companies are incorporated take enforcement or investigative positions informed by the fiduciary duty arguments raised in this article. OpenAI's ongoing conversion from a capped-profit to a full for-profit structure is a near-term trigger event that could test exactly the accountability mechanisms the article identifies as deficient, and any regulatory challenge or board dispute arising from that conversion will have immediate vendor risk implications. The SEC's evolving AI governance guidance and investor disclosure expectations are also likely to reference corporate structure and fiduciary duty questions as the frontier AI sector matures, making this legal analysis a useful framing document for ESG and investor disclosure programs.