Artificial intelligence has moved from curiosity to commonplace in legal practice.
Today, lawyers routinely use AI to draft pleadings, summarise evidence, review contracts, analyse discovery, conduct legal research and prepare client advice. The question is no longer whether AI belongs in legal practice. It already does.
The real question is whether the legal profession has developed the governance framework needed to use it responsibly.
Recent court decisions suggest the answer is no.
Over the past two years, courts around the world have sanctioned lawyers for relying on AI-generated authorities that never existed. While fabricated citations have captured headlines, they are merely the most visible symptom of a much broader problem.
The real issue is not hallucinated case law.
It is the mistaken belief that professional judgment can be outsourced to software.
Existing ethical duties already apply
Lawyers already owe duties of competence, confidentiality, independence, candour to the court and proper supervision. None of those obligations disappear because AI is involved.
If anything, AI raises the standard.
The lawyer remains responsible for every document filed, every piece of advice given and every strategic decision made. AI may generate content, but it cannot assume professional responsibility.
Ethics cannot be delegated.
The illusion of competence
One of generative AI’s greatest strengths is also its greatest danger.
Its output is fluent, persuasive and professionally presented. It often sounds authoritative even when it is incomplete, misleading or entirely wrong.
That creates what might be called an “illusion of competence”.
Experienced practitioners understand that legal work is not simply producing words. It requires identifying relevant facts, weighing competing evidence, recognising uncertainty, exercising discretion and anticipating legal risk.
Those are fundamentally human judgments.
Large language models can assist with reasoning, but they cannot replace professional judgment.
The governance gap
The greater risk facing the legal profession is not rogue lawyers using AI irresponsibly.
It is well-intentioned lawyers using AI without adequate governance.
Many firms still lack clear policies addressing:
- verification of AI-generated legal authorities;
- protection of confidential client information;
- supervision of junior practitioners using AI;
- disclosure of AI-assisted work where appropriate;
- billing practices for AI-assisted legal services;
- quality assurance and human review; and
- audit trails demonstrating how legal conclusions were reached.
General guidance to “be careful” is no longer sufficient.
As AI becomes embedded in everyday legal workflows, firms need structured governance comparable to their trust accounting, cybersecurity and conflict-management systems.
From AI competence to an AI duty of care
The legal profession should begin thinking beyond AI competence towards an AI duty of care.
An AI duty of care recognises that lawyers who deploy AI owe clients a positive obligation to ensure its use is reasonable, proportionate and appropriately supervised.
That duty extends beyond verifying citations.
It includes ensuring that:
- confidential information is handled securely;
- AI outputs are independently verified;
- material facts have not been omitted;
- reasoning can be explained;
- significant decisions remain human decisions; and
- clients are not exposed to foreseeable risks created by automated systems.
In other words, lawyers must exercise the same level of professional care when supervising AI that they exercise when supervising junior lawyers.
The next wave of professional liability
Fake citations may become increasingly rare as AI systems improve.
More subtle failures are likely to replace them.
A medical chronology that omits a crucial diagnosis.
A contract review that overlooks an unusual indemnity clause.
An AI-powered client intake system that unintentionally filters out legitimate claims.
Sensitive client information uploaded into insecure public platforms.
Clients charged partner rates for work performed largely by automated systems.
These failures will not necessarily involve spectacular misconduct.
They will involve inadequate supervision.
And that is ultimately an issue of professional responsibility.
Verification is becoming the profession’s competitive advantage
As AI becomes ubiquitous, the value lawyers provide will shift.
Clients will increasingly assume that AI assists with drafting, summarisation and research.
What they will pay for is verification, judgment and accountability.
The lawyer’s role evolves from being the primary producer of legal text to becoming the verifier of legal reasoning.
That requires more than checking citations.
It requires verifying facts, testing assumptions, evaluating evidence and explaining why a conclusion is legally defensible.
The future of legal practice will belong not to the firms using the most AI, but to the firms best able to demonstrate that their AI-assisted work can be trusted.
Trust remains the profession’s defining obligation
The legal profession does not need to prohibit AI.
Nor should it.
AI has enormous potential to improve efficiency, increase access to justice and reduce the cost of legal services.
But those benefits will only be realised if the profession develops governance frameworks that place human accountability at the centre of AI-assisted practice.
Lawyers remain fiduciaries.
Clients retain lawyers because they trust their judgment—not their software.
At the end of every AI-assisted matter, only one name appears on the advice, pleading or affidavit.
That name belongs to the lawyer.
And with it comes an obligation that no algorithm can assume: the duty to exercise independent professional judgment with reasonable care.
In the AI era, that is what an AI duty of care should mean.
source: Above the Law
Leave a Reply