The Strange New Skill Lawyers Need Is Knowing When Not to Trust AI
AI tools can draft a memo in minutes, summarize a stack of documents before lunch, and brainstorm arguments faster than any associate. For busy lawyers, that speed is incredibly appealing. Who wouldn't want hours of tedious work handled in moments? Tempting, for sure.
But there's a catch. The same tools that write so smoothly can also produce answers that sound completely confident and are completely wrong. Some have generated case citations that look real but point to decisions that never existed. That's a real problem.
That creates a strange new paradox for the legal profession. AI can make legal work faster and more efficient, but only for lawyers who know when to trust it, when to double-check it, and when to set it aside entirely. Judgment is the new skill.
Why AI Can Sound Certain Even When It's Wrong
Generative AI models work by predicting likely sequences of words based on patterns in their training. They're remarkably good at producing fluent, professional-sounding text. But fluency isn't the same as accuracy, and the model doesn't automatically know when it's wrong. It predicts rather than knows.
That's why attorneys exploring Claude for lawyers or any other AI tool need to independently verify legal research and factual claims. A polished paragraph can still contain an outdated rule, a misstated holding, or a detail that simply isn't true. Verification isn't optional.
The tone can be misleading too. AI tools rarely hedge the way a careful human researcher might. Without visible uncertainty, it's easy to assume an answer is reliable. Lawyers have to supply that skepticism themselves, every single time. Healthy doubt is essential.
The Strange Problem of AI-Generated Legal Citations
Fabricated citations are one of the most notorious AI problems in law. In a widely reported 2023 federal case in New York, Mata v. Avianca, attorneys were sanctioned after submitting a brief citing nonexistent cases generated by an AI chatbot. It was a costly lesson.
Fake citations aren't the only risk. AI can also cite real cases but mischaracterize what they hold, attribute quotes that don't appear in the opinion, or rely on decisions that have since been overturned. Those errors can be much harder to spot.
That's why checking primary authorities remains essential. Reading the actual opinion, confirming it's still good law, and verifying every quotation protects clients and courts. Many courts have also adopted rules or standing orders addressing the use of AI in filings. Diligence still rules.
Where AI Can Actually Help Legal Professionals
Used carefully, AI can be a genuinely useful assistant. It can help organize large sets of documents, pull key dates from records, or summarize long transcripts so attorneys can focus on what matters most. Those tasks can eat up hours of billable time.
Brainstorming is another strong use. AI can suggest arguments, counterarguments, or questions to explore, giving lawyers a starting point for deeper analysis. It can also help draft routine correspondence, outlines, or first versions of documents. It's a helpful thinking partner, not a final authority.
The real key is human oversight. AI output works best as a draft or a starting point, not a final product. Lawyers who review, edit, and verify everything before relying on it can gain efficiency without sacrificing accuracy or judgment. Oversight remains non-negotiable.
Why Confidentiality and Professional Judgment Still Matter
Client confidentiality is a core professional duty. Before entering sensitive information into any AI tool, attorneys need to know how that data is handled, stored, and used, and whether the tool's terms align with their ethical obligations. Due diligence comes before any upload.
The American Bar Association addressed these issues in Formal Opinion 512 in 2024, emphasizing duties like competence, confidentiality, communication, supervision, and reasonable fees when using generative AI. Many state bars have issued their own guidance as well. Local rules may add even more.
Professional judgment can't be outsourced. Lawyers remain responsible for the advice they give and the documents they file. Evaluating tools carefully, setting clear internal policies, and training staff on appropriate use are all part of using AI responsibly. Accountability stays with the lawyer, always.
Conclusion
AI can make legal work faster and more efficient, but it can't replace careful human thinking. Convincing answers can still contain errors, invented authorities, or misstatements that create real problems for clients and lawyers alike. Speed without accuracy can backfire, sometimes very publicly.
The most valuable skill isn't knowing how to use AI. It's knowing when not to trust it. Verifying citations, checking facts, and reviewing output critically turn AI from a risk into a genuinely useful tool. That skill separates careful lawyers from careless ones.
Responsible use also means protecting confidentiality and maintaining professional judgment. Lawyers who combine AI's speed with human oversight and healthy skepticism are best positioned to benefit from these tools without being burned by them. The future likely belongs to lawyers who stay thoughtfully skeptical.
