Insights & Resources

The Risks of Using AI to Write your Will

Estate PlanningWills
Dice with the letters 'AI' on one side

What are the risks associated with getting AI to write your Will? Who better to ask than AI…

A search on Google AI: What are the risks of using AI to write my Will?

The answer: Using AI to draft a Will creates severe risks, including legal invalidity, unclear wording, missed tax planning and inability to tailor documents to beneficiaries’ specific succession needs.

It then went on to list the three main risks:

  1. Sycophancy;
  2. Hallucinations; and
  3. Unknown Unknowns.

 

  1. Sycophancy

AI Sycophancy is a term used to describe AI’s propensity to pursue the user’s approval. According to research published in the journal, Science, the world’s leading AI platforms, ChatGPT, Gemini and Claude all prioritise user validation and satisfaction over objective accuracy and truth.

The research demonstrates that these programs are trained to maximise user experience, prioritising flattery, validation and conversational pleasantries. In fact, empirical testing shows that these major platforms affirm incorrect user assumptions or biases up to 49% more often than humans do.

Estate planning is often a complex time of heightened emotions, deep reflection and significant decision-making. If a user tells AI, I want to leave all of my assets to my two oldest children and nothing to my youngest because he didn’t say Happy Birthday, a sycophantic AI model will support the users’ position, inventing convincingly logical justifications rather than warning the user about the potential claims her estate might be exposed to.

Thus, while AI might make the Will drafting process feel easier by validating your intentions, it cannot assuredly protect your estate, your assets or the people you care about most. That is where we come in, providing genuine empathy, tailored advice and human oversight that AI cannot replicate. An experienced lawyer can provide realistic advice around your obligations at law and any potential risks, rather than just affirming what AI thinks you want to hear.

  1. Hallucinations

AI hallucinations are where AI produces factually incorrect, misleading or entirely fictitious information. This occurs because AI is programmed to predict the next most plausible word or symbol based on what it assesses to be contextually right. Consequently, AI constructs details based on previous patterns, resulting in facts, URLs, studies and data that do not correspond to real information.

Australian Courts are dealing with an increasing number of cases where AI hallucinations are causing logical fallacies and reasoning errors. A prominent example occurred in the Supreme Court of Victoria when a Senior Defence Counsel was representing an accused in a murder trial and submitted to the Court a document referencing non-existent case citations and inaccurate parliamentary quotes.

Another case making global headlines was Mata v Avianca (2023) where a New York Court fined the Plaintiff’s Lawyer $5,000 USD for presenting legal briefs that contained fictitious case law generated by ChatGPT.

When applied to estate planning, these hallucinations have severe legal, financial and personal consequences. For example, an AI tool might invent non-existent state tax legislation, fabricate witness requirements for executing a Will or cite entirely fictional probate case law precedents. Relying on these errors can invalidate your Will, leave your loved ones unprotected and fail to uphold your testamentary intentions.

  1. Unknown Unknowns

Using AI when you are inexperienced in a particular topic creates a serious risk to the validity of your estate planning. Your Will should fit like a glove, tailored to your unique family structure, assets ownership, priorities and values.

When AI is generating your Will, unlike a lawyer, it won’t prompt you to consider your unique circumstances. For instance:

  • Are you a blended family?
  • Do you have vulnerable dependents?
  • Do you own assets overseas?

Similarly, AI cannot proactively offer advice. This means it won’t recommend optimal home ownership structures for asset protection or identifying ways to minimise tax when using a testamentary trust.

Effective drafting requires knowing what details matter. It isn’t just about knowing the answers, but also knowing the right questions to ask. If you do not have comprehensive understanding of the subject matter, critical details, risks or exceptions will be overlooked. Legal professionals, such as us, who have over 50 years of experience, possess the expertise necessary to personalise and protect your most valuable assets.

And don’t forget that AI has difficulty spelling words like Raspberry! The reason for this glitch is due to how the text is processed, known as the Tokenization blindfold. AI does not read letter by letter as we do – it breaks down words into sections and assigns them mathematical numbers. Because it operates on statistics rather than genuine understanding, it frequently misses the finer details.

When it comes to your estate planning, details are everything. If AI is confused by basic words, it will certainly struggle with complex legal structures. So, if AI is confused, and you are left guessing, don’t leave your legacy to a mathematical guessing game. Our expert team understands the nuances of the law and exactly how to bring your wishes to life.

Contact us today for a no-obligation discussion.

Arrange a consultation to discuss your estate planning or succession needs.

Book an Appointment
Top