From Black Boxes to Best Practices: AI Ethics For Estate-Planning AttorneysFrom Black Boxes to Best Practices: AI Ethics For Estate-Planning Attorneys

Artificial intelligence (AI) transforms estate and trust law, providing tools that streamline drafting, research and dispute resolution. Attorneys can use AI to analyze complex trust provisions, draft wills or evaluate fiduciary performance. However, these technologies introduce ethical challenges, particularly concerning client confidentiality. Rule 1.6 of the Model Rules of Professional Conduct requires attorneys to safeguard client information, a principle that can be jeopardized if AI tools are used without proper precautions.

I’ll explore the risks AI poses to confidentiality and outline the ethical obligations attorneys must uphold. In my next column, I’ll provide actionable steps to use AI responsibly in an estate-planning practice.

The Confidentiality Dilemma

General-purpose AI tools, particularly free platforms, aren’t tailored to the stringent confidentiality needs of legal practice. General-purpose AI tools are defined as AI systems designed to perform various tasks across multiple industries rather than being tailored to a specific domain or use case. These tools are often user-friendly and accessible to the general public, making them versatile for a range of applications. However, because they aren’t specialized, they may lack the privacy safeguards and compliance mechanisms required for sensitive industries like legal practice.

General-purpose AI tools typically include features such as:

• Natural language processing (for example, generating text and answering questions)

• Image and video recognition

• Predictive analytics

• Data summarization and trend analysis

General-purpose AI tools are highly adaptable but not inherently equipped to handle the strict confidentiality requirements of estate-planning attorneys or other legal professionals. Estate-planning attorneys handle sensitive details, such as family relationships, trust structures and tax strategies, which can be exposed if processed through unsecured AI systems. See “Examples of General-Purpose AI Tools,” this page, for tools that may not meet confidentiality requirements. 

Data Retention 

Many AI platforms retain user inputs to refine their algorithms, which may result in sensitive data being stored indefinitely. This poses significant risks for estate-planning attorneys.

Example: An attorney uses a free AI tool to draft a cover letter for a high-net-worth client’s estate-planning documents. The input includes the client’s asset structure, intended charitable bequests and provisions for a blended family. The AI system retains the data for training purposes, potentially exposing the client’s private information to unintended parties through other outputs.

Limited Transparency

Many AI tools operate as “black boxes,” meaning their internal workings, data processing methods and storage practices are opaque to users. These tools often lack transparency regarding where user inputs are sent, how long they’re stored and who might have access to the data. For estate-planning attorneys, this ambiguity poses a significant risk, as the highly sensitive nature of client information, such as the structure of an irrevocable trust, details of cross-border assets or tax strategies, requires assurance that such data is handled securely. 

Without clear documentation or guarantees from AI providers, attorneys can’t verify whether the systems they rely on meet the stringent confidentiality and compliance standards demanded by the profession. This lack of visibility raises questions about potential vulnerabilities, such as unencrypted transmissions, insufficient access controls or compliance with privacy regulations like the European Union’s General Data Protection Regulation (GDPR).

Example: A trust attorney uses AI to evaluate tax implications for a foreign trust. The platform processes the data on international servers, but the attorney has no visibility into whether the data is encrypted or accessible to unauthorized parties. As a result, sensitive financial details, such as the value of offshore accounts or the identity of beneficiaries, could be at risk of exposure, violating the attorney’s ethical obligations and potentially leading to client harm.

Jurisdictional Complexities

Estate planning often involves clients with cross-border assets, international beneficiaries or global investments, making the secure handling of sensitive data a critical concern. When AI tools process data across jurisdictions, attorneys must consider the privacy laws and data protection regulations governing each region where the data may be transmitted, processed or stored. For example, GDPR imposes strict requirements on the handling of personal data, including consent protocols, data minimization and specific safeguards for data transfers to non-compliant countries. If an AI tool processes client data on servers located in jurisdictions with lax privacy protection or insufficient regulatory oversight, it can expose both the attorney and the client to significant risks. These include unauthorized access to sensitive details, non-compliance with international privacy laws and reputational harm to the law firm.

The risks are compounded by the opaque nature of some AI platforms, which may not disclose the exact locations of their data centers or the specific protocols used to protect cross-border data transfers. Failure to verify these details could inadvertently lead to ethical breaches, regulatory violations or the exposure of critical financial and personal information. Attorneys must proactively ensure that AI tools comply with all applicable privacy laws, especially when handling sensitive client matters such as international estate plans, offshore trusts or multi-jurisdictional inheritance disputes.

Example: An attorney drafting an international estate plan for a client with significant cross-border assets uses an AI tool to evaluate potential tax efficiencies and compliance strategies. Unknown to the attorney, the AI platform processes data on servers in a jurisdiction that doesn’t comply with GDPR or similar data protection frameworks. As a result, details such as the client’s offshore account balances, real estate holdings and beneficiary identities may be exposed to unauthorized access or misuse. If this mishandling comes to light, the client could face financial repercussions, and the attorney or firm may face legal and regulatory exposure for failing to ensure proper safeguards. This scenario highlights the importance of vetting AI tools thoroughly to confirm their adherence to international privacy standards and secure data handling practices.

Key AI Features 

To use AI responsibly, estate-planning attorneys must understand how these systems handle data and the measures that can mitigate risks to client confidentiality. The methods AI tools use to process, store and analyze data vary widely, and not all are suitable for the high standards of privacy and security required in legal practice. By prioritizing key features such as data isolation, on-premises processing, end-to-end encryption and federated learning, attorneys can select tools that safeguard sensitive information while ensuring compliance with ethical and legal obligations.

Data isolation. Data isolation ensures that each user’s input is processed in a secure, independent environment, preventing it from being stored, reused or incorporated into the AI system’s broader training models. This feature is particularly important for estate-planning attorneys, as it guarantees that client data, such as asset lists, medical information or trust provisions, remains private and can’t be inadvertently exposed to other users or retained indefinitely by the AI provider.

Example: A law firm specializing in estate planning uses an AI tool with strict data isolation to draft special needs trusts for clients. In one case, the attorney inputs details about a client’s assets, the medical condition of a dependent and specific language outlining the terms of the trust. Because the tool operates in an isolated environment, the information is processed only for that specific task and is deleted immediately afterward. Unlike general-purpose AI platforms that might retain input data to refine their algorithms, the data isolation feature ensures that the client’s sensitive details, such as the client’s medical condition or inheritance structure, can’t be used in future queries or seen by other users. This assures both the attorney and the client that confidentiality is maintained.

 On-premises processing. On-premises AI tools allow organizations to deploy and run AI systems locally within their secure networks. By eliminating the need to transmit data to external servers, this approach ensures that sensitive client information never leaves the firm’s controlled environment. For estate-planning attorneys handling high stakes matters like dynasty trusts or multi-jurisdictional estate plans, on-premises processing significantly reduces the risk of unauthorized access or external data breaches.

Example: An attorney working on a dynasty trust for a client with substantial assets across multiple states uses a locally hosted AI platform to draft and refine trust provisions. The platform operates entirely within the law firm’s secure network, ensuring that sensitive details, such as the trust’s terms, generational asset allocation strategies and tax minimization clauses, are never transmitted to external servers. This setup not only safeguards against hacking and unauthorized access but also allows the firm to comply with strict data protection regulations that may apply to the client’s information. Additionally, the firm’s IT team can monitor the tool’s operations in real time, providing an added layer of security and control. 

End-to-End Encryption

End-to-end encryption secures data during its entire lifecycle—both in transit and at rest. This ensures that only authorized users can access the information, even if it’s intercepted during transmission or stored on a server. For estate-planning attorneys, encryption is critical when using AI tools to analyze sensitive matters such as fiduciary disputes, as it prevents third parties from accessing confidential communications or client details.

Example: An attorney representing a client in a fiduciary dispute uses an AI-powered analysis tool to identify patterns and precedents that support their case. The tool encrypts all data the attorney enters, including claims of mismanagement by the trustee, financial account statements and communications among beneficiaries. Even if a hacker intercepts the data during transmission or gains access to the storage server, the encryption ensures that the information can’t be decrypted or understood without the proper credentials. This robust security measure not only protects the client’s sensitive information but also reinforces the attorney’s ethical duty to maintain confidentiality.

Federated Learning

Federated learning is an advanced AI technique that trains machine learning models across decentralized devices or servers while keeping raw data local. This approach enhances security by eliminating the need to transmit sensitive information to a central server. For estate-planing attorneys, federated learning is particularly useful for tasks like document review or trust analysis when preserving client confidentiality is paramount.

Example: A law firm implements federated learning technology for its AI-powered document review system, which analyzes complex trust documents for inconsistencies or opportunities for optimization. For example, the system examines provisions in an irrevocable trust to identify potential conflicts with tax laws or missed opportunities for asset protection. Unlike traditional AI systems that upload the entire document to a central server for processing, the federated learning model processes each document locally on the firm’s secure network. This ensures that sensitive client details, such as the identities of beneficiaries or the values of specific trust assets, remain protected. The AI system still improves its capabilities by analyzing aggregate data trends across the firm’s network without exposing individual client information externally.