AI is rapidly embedding itself into trust administration and family offices, while states in the US are loosening fiduciary protections to encourage competition. That presents challenges for UHNW families and creates risks, the author argues.

All the noise and commotion around AI – such as its impact on wealth managers’ business and prospects – can mask other, perhaps more subtle implications that this technology has. An example is the trusts sector and how those who advise clients about protecting wealth go about their roles.

To discuss this is one of our regular writers, lawyer Matthew Erskine, who is also a member of our editorial board. As ever, if you value this content, please comment and enter the conversation, and if you want to push back, suggest ideas or topics, please do so. The usual editorial disclaimers apply to guest writers’ views. Email tom.burroughes@wealthbriefing.com and amanda.cheesley@clearviewpublishing.com
 

A quiet collision is underway in the world of multigenerational wealth preservation, and most families – and their advisors – have not yet recognized its implications.

On one side, artificial intelligence is rapidly embedding itself into the infrastructure of trust administration and family office operations. The UBS Global Family Office Report 2025 found that more than 80 per cent of family offices expect to invest in AI within two to three years, with applications spanning financial reporting, portfolio analytics, risk monitoring, and investment decision support. The RBC and Campden Wealth North America Family Office Report 2025 reported that family offices have tripled their use of AI for operational purposes compared with the prior year. AI is no longer a curiosity at the margins – it is becoming the operational backbone of how trustees and family offices manage wealth.

On the other side, state legislatures are engaged in a decades-long competition to attract trust business by systematically relaxing fiduciary protections. States including South Dakota, Delaware, Tennessee, and New Hampshire have enacted provisions that allow settlors to override the duty of good faith, create perpetually silent trusts where beneficiaries may not even know they are beneficiaries, insulate trustees from virtually all liability through expansive exculpatory clauses, and establish directed trust structures in which no single party bears clear fiduciary accountability. These developments are well-documented among trust law specialists, though they have received comparatively little attention from the families whose wealth they are designed to attract.

Each of these trends, considered independently, poses significant challenges for ultra-high net worth families. Together, they create something far more concerning.

The convergence problem
When a trustee deploys an AI-driven investment platform inside a trust structure that has eliminated the duty of good faith, waived beneficiary disclosure, capped trustee liability at willful misconduct, and designated a non-fiduciary trust director for investment decisions, a fundamental question arises: who is accountable when something goes wrong?

The trust director, in many jurisdictions, is not a fiduciary. The directed trustee is exculpated for following the director’s instructions. The AI platform is a tool, not a legal person. And the beneficiaries may not know the trust exists, may have no right to information about how decisions are being made, and may have no practical avenue for recourse.

This is what I call the fiduciary vacuum – a structural gap in which AI-powered decision-making operates with diminishing human oversight inside trust frameworks that simultaneously eliminate meaningful accountability for the consequences.

Why UHNW families are uniquely exposed
The fiduciary vacuum creates distinct vulnerabilities depending on the nature of the assets involved.

Family enterprises embody values, relationships, and intergenerational commitments that resist algorithmic optimization. An AI system evaluating a family manufacturing company through purely financial metrics might recommend liquidation or sale to a private equity buyer – a decision that could be financially defensible but devastating to the family’s legacy, its employees, and its community standing. Within an eroded trust structure, a non-fiduciary trust director following an AI recommendation may face no duty to weigh these non-financial considerations.

Art, collectibles, and emotionally significant assets require stewardship that balances financial returns with aesthetic vision, provenance integrity, and cultural meaning. An AI system optimizing portfolio efficiency might recommend breaking up a coherent collection without accounting for its value as an integrated whole. In a silent trust with eroded protections, beneficiaries may learn of the liquidation only after the damage is irreversible.

Multi-jurisdictional complexity compounds the problem. AI tools may not account for the interaction between different trust laws, tax regimes, and fiduciary standards across borders. The ability of trustees to change a trust’s governing law and decant to a more permissive jurisdiction means that protections originally built into the trust instrument can be dismantled – potentially without any beneficiary’s knowledge.

The behavioral dimension
The fiduciary vacuum is made worse by well-documented behavioral patterns. Research on automation bias consistently shows that humans defer to automated systems even when those systems produce questionable results. Trustees who rely on AI tools are subject to this tendency, particularly when the trust structure itself removes the consequences of insufficient scrutiny.

The “human-in-the-loop” safeguards that underpin current regulatory guidance – including the American Bar Association’s Formal Opinion 512 on generative AI and the SEC’s fiduciary framework – assume that a human can meaningfully evaluate an AI system’s output. When the AI is processing thousands of data points and making portfolio adjustments at a speed that renders genuine human review impractical, the human-in-the-loop becomes a legal fiction. And when the trust structure has eliminated the consequences of failing to supervise, fiction becomes a dangerous reality.

What families should ask
The fiduciary vacuum is not an abstract policy debate. It is a present reality that affects trusts being drafted and administered today. Families and their advisors should be asking hard questions:

Does our trust’s governing law permit the elimination of the duty of good faith? If so, has our trust instrument done so?

Do our exculpatory clauses insulate the trustee from liability for failures related to AI oversight?

If our trust is silent, do beneficiaries – or at minimum, a designated representative with genuine fiduciary duties – have access to information about how AI tools are being used in trust administration?

If our trust is a directed trust, does the trust director bear meaningful fiduciary responsibility for AI tool selection and oversight?

Could our trust be decanted or moved to a jurisdiction that would weaken existing protections?

These are questions that most estate planners are not yet asking, because the convergence of AI adoption and fiduciary erosion has not yet been framed as a single, integrated problem. That framing is overdue. The families who address the fiduciary vacuum proactively – before a crisis reveals the gap – will be the ones whose trusts, enterprises, and legacies emerge intact. The tools for doing so exist. What has been missing is the recognition that they are needed.

Ábout the author

Matthew F Erskine (pictured above) is an estate planning attorney at Erskine & Erskine specializing in sophisticated planning strategies for ultra-high net worth families, family enterprises, and family offices.