Writing for the AI Audience: A New Legal Skill

Lawyers, it’s time to learn how to write with AI as the main audience.

There’s a lot of talk about lawyers learning how to use AI. But an equally important skill is learning how to write for AI.

Not long ago, the first reader of a client memo, legal survey, or contract draft was your client, a manager, or a law firm partner.

Today, the first reader is often an AI tool they’re using to summarize, analyze, or critique the work.

3 Moments That Brought This Home

1️⃣ AI is the new TLDR

"I turned your benchmarking report into a podcast with AI."

More than one person has told me this lately.
They “read” the legal AI benchmarking reports by asking an AI tool to summarize them, then either listened to the summary or skimmed it.

That’s often where their engagement stopped.

Unless they asked follow-up questions, all the nuance, the footnotes, caveats, and qualifiers vanished at the point of summarization.

2️⃣ AI is the new risk gatekeeper

"Before I review a contract from the counterparty, I always run it through my enterprise AI tool to highlight key risks and flag unfavorable terms."

A senior counsel told me this recently.
The AI makes the first pass, shaping his perception of the document before he even reads it.

3️⃣ AI is our new critic

A law firm partner recently told his associates:

“Don’t hand in a work product that’s worse than ChatGPT.”

That’s how one law-firm partner now sets expectations for associates.

Clients, partners, and business teams already use AI to check legal work for gaps, inconsistencies, or tone.

AI isn’t just assisting lawyers to generate work; it’s also being used to grade a lawyer's work.

How AI May Get Lost in Translation

Our Legal AI Benchmarking research confirms that when AI tools read legal text, meaning may get lost and subtle but consequential details may get dropped through omissions, misplaced context, or the quiet removal of qualifiers.

For a detailed discussion of these failure modes, see Section 3.4 of the Info Extraction Benchmark Report (“AI Assistant Weaknesses”) and Section 4.2 of the Contract Drafting Benchmark Report (“Humans vs. AI: Strengths and Weaknesses”).

Designing Legal Writing for the Age of AI

As AI increasingly reads before humans do, a lawyer’s writing now has to serve two audiences at once: the machine that processes it and the human who relies on its output.

Here are three ways to design legal writing that remains clear and accurate even after it passes through an AI lens.

1. Self-Audit With the Tools Your Readers Use

What to do: Run your draft through an AI tool your reader might use.What to look for: Does the summary capture your main points? Do any disclaimers or qualifiers disappear?

2. Include Critical Details in the Main Text

What to do: Assume footnotes, pop-ups, and side comments won’t make it into the AI summary. Keep essential qualifiers in the body text.

Example: Instead of writing “The exemption applies (see footnote 4),” write “The exemption applies only if the transaction is below SGD 5 million.”

That way, the condition stays visible even in an AI summary.

3. Use Consistent Terms and Standalone Clauses

What to do: AI tools currently index and summarize in chunks, so clarity and consistency matter more than elegant phrasing. Use consistent terms, clear headings, and minimal cross-referencing.

Example: Instead of saying, “See section 7 for supplier obligations,” restate it briefly: “The supplier must maintain audit logs (see section 7 for retention details).”

That way, even if the document is split or summarized, the meaning remains intact.

Conclusion

Your first reader may soon no longer be human.

Writing for AI doesn’t mean writing like a machine; it means writing so that meaning, nuance, and risk survive the interpretation of the machine.