Courts across the country are adopting rules that require lawyers to disclose when they "use" AI in their filings. The intent behind these rules is understandable. But there is a word doing a lot of heavy lifting in that sentence, and it deserves more scrutiny: "use."

Consider a few scenarios.

I pull up a legal-research platform that happens to be powered by AI. I feed it my adversary's brief and ask it to identify every case cited so I can pull and review each one myself. Did I "use" AI? The tool saved me ten minutes of skimming footnotes, but every word of analysis is mine.

Now suppose I go a step further. I ask the platform to generate short summaries of each cited case so I can triage which ones warrant a close read. I still read the key cases in full. I still do the analysis. But AI helped me decide where to focus. Must I file a disclosure?

Or take a different scenario entirely. I draft a brief from scratch — every argument, every citation, every piece of analysis is my own work. But the brief is two pages over the limit. I run it through an AI writing tool to tighten the prose, cut redundancies, and bring it within the page count. The substance is untouched. Is a disclosure required?

These are not hypotheticals I am inventing to be difficult. These are questions that lawyers are confronting right now, and the current patchwork of court rules does not answer them clearly. The word "use" is doing too much work without enough definition. A lawyer who asks AI to draft an entire brief from whole cloth is "using" AI. A lawyer who asks AI to fix a typo is also "using" AI. Surely those two things are not the same, and surely they should not trigger the same disclosure obligations.

We are all still learning — learning which tools work, which tasks are well-suited to AI, and where the guardrails need to be. That learning is happening in real time, and the rules are struggling to keep pace.

I'd love to hear from this community.

Clients: How do you expect your outside counsel to use AI? Which uses do you prohibit? Discourage?

Practitioners: How are you using AI in your practice? Which tasks have you found it's best suited for? Where do you draw the line on disclosure? And do you think the current rules are getting it right — or do they need to catch up to how lawyers are actually working?

Drop your thoughts in the comments. 👇