AI Is a Great Tool. It Is a Terrible General Counsel.

Artificial intelligence has become remarkably good at drafting contracts, summarizing documents, preparing demand letters, and answering legal questions. It is fast, inexpensive, and available at 11:30 p.m. when your lawyer may not be.

That does not mean it should be your lawyer.

For small and mid-sized businesses in particular, the temptation is understandable. Why pay an attorney to review an agreement when an AI platform can produce a five-page analysis in thirty seconds? Why call counsel about a demand letter when AI can draft a response that sounds appropriately indignant—complete with several impressive-looking citations to cases that may or may not exist?

Sometimes, that works just fine.

Sometimes, it confidently makes things up.

And that distinction can get expensive.

The Problem Is Not That AI Is Bad

I use AI regularly. Most lawyers do now, whether they advertise it or not.

Used correctly, it can be an excellent tool. It can help organize complicated facts, identify issues for further review, compare contractual language, summarize lengthy documents, generate first drafts, and make lawyers more efficient.

The important words there are “used correctly.”

AI does not actually understand your business, your deal, your litigation posture, or the consequences of getting something wrong. It predicts what a good answer should look like based on patterns it has identified in enormous amounts of data. Usually, the result is impressive.

Occasionally, it is impressively wrong.

AI systems still “hallucinate”—the polite technological term for confidently fabricating information. They can cite cases that do not exist, misstate statutes, overlook exceptions, misunderstand defined terms in agreements, or recommend contractual language that sounds sophisticated but does not accomplish what the parties actually intend.

Worse, the output does not come with a flashing red light that says:

WARNING: I JUST MADE THIS PART UP.

It reads every bit as authoritative as the correct answer. Sometimes more so. (Who isn’t more authoritative when they’re not bound by the truth?)

The Real Danger with AI Is Not Knowing What You Don’t Know

Lawyers make mistakes too. The difference is that an experienced lawyer reviewing an AI-generated answer usually has enough background to recognize when something looks wrong.

A business owner may not.

If AI tells me that Virginia law requires a particular result, I generally have a sense of whether that proposition passes the smell test. I know what statutes to check, what cases matter, what exceptions might apply, and when an answer is suspicious enough to warrant digging deeper.

If you do not regularly practice in that area of law, you may have no reason to question it.

That is particularly dangerous because the mistakes tend to matter most where the legal analysis is least obvious.

A contract provision can look perfectly reasonable while quietly shifting millions of dollars of risk.

A demand letter can sound aggressive while accidentally admitting something harmful.

A lease provision can appear “standard” while eliminating a right your business assumed it had.

A corporate document can contain every impressive phrase AI has ever encountered and still fail to accomplish the transaction the parties intended. It will, however, be beautifully formatted.

And by the time someone realizes it, the contract may already be signed, the deadline may have passed, or the email may be Exhibit A.

“Can AI Review This Contract?” Is the Wrong Question

Of course it can.

The better question is: Who is reviewing AI?

Businesses should absolutely use technology to operate more efficiently. They should also expect their lawyers to use it. AI can reduce the amount of time attorneys spend on tasks that do not require much judgment and allow us to focus more attention on the issues that do.

That is a good thing for clients.

But there is an enormous difference between a lawyer using AI as a tool and a business using AI as a substitute for a lawyer.

One involves technology assisting professional judgment.

The other involves hoping the robot noticed the indemnification provision before you signed it.

There Is a Sensible Middle Ground with AI

Businesses do not need to send every routine email or minor agreement to outside counsel. Nor should companies be afraid to use AI.

Use it to brainstorm. Use it to summarize. Use it to create a first draft. Use it to identify questions. Use it to help explain a document before you send it to counsel.

There is also the (not so) small matter of confidentiality.  Feeding contracts, privileged communications, and sensitive business information into an AI platform raises an entirely different set of issues – ones that businesses should fully understand before treating AI like another member of the team, and ones I’ll tackle in a follow-up post.

But when the answer affects meaningful money, liability, ownership, litigation, employment rights, real estate, regulatory compliance, or another material business risk, having an actual lawyer review the result remains a very good investment.

AI is getting better quickly.

It may someday replace lawyers altogether.

Until then, I would probably have a lawyer review that clause. Preferably one who bills by the hour and not the token.

An AI-generated contract provision, demand letter, or business document may look correct on its face, but overlooked risks, inaccurate legal analysis, or poorly drafted language can lead to costly disputes down the road. Sands Anderson's Business Litigation attorneys help businesses evaluate risk, resolve conflicts, and protect their interests when high-stakes disputes arise.