Courts Are Starting to Pick AI Tool Winners: Breaking Down Morgan v. V2X Inc.

7 minutes learn

Revealed Apr 17, 2026

The details behind Morgan v. V2X Inc.

The defendant, V2X, sought a protecting order earlier than discovery even started. They requested the court docket to require sure protections over data labeled by events as confidential. These protections included deletion necessities and accountability measures for the plaintiff.

The court docket granted a typical protecting order. If data is labeled confidential, deal with it with the agreed diploma of safety. However then the decide seen the plaintiff was utilizing AI to help with litigation preparation. The defendant requested for the protecting order to be amended to exclude the plaintiff’s use of AI instruments solely.

The plaintiff pushed again, arguing amongst different issues that disclosing which AI instruments they used might itself reveal litigation technique. Choose Braswell cut up the distinction, and the ensuing order addresses three questions that matter for each lawyer utilizing AI in litigation. 

Professional se litigants and AI work product safety

The primary query was easy. Can a professional se litigant’s use of generative AI qualify for work product safety?

Choose Braswell stated sure. She cited the Heppner ruling out of the Southern District of New York and Warner out of Michigan, each of which we’ve mentioned earlier than, and carried their reasoning ahead into this circuit.

Her reasoning is value studying carefully. The court docket wrote that “The significance of making use of these [work product] protections to professional se litigants is magnified within the context of AI—one of the vital highly effective information instruments ever to turn out to be accessible to the lots. It’s because professional se litigants are compelled to behave as each occasion and advocate, concurrently. And for the primary time in historical past, widespread entry to highly effective expertise could make that twin position,” which means the professional se litigant appearing as each occasion and lawyer, “surmountable.” As a result of Rule 26(b)(3) of the Federal Guidelines of Civil Process, which governs work product safety, doesn’t “situation work product safety over AI supplies on the involvement of counsel… ”

In plainer phrases, denying work product safety to self-represented litigants just because they used AI as a substitute of a lawyer is, on this decide’s view, unconscionable. AI is without doubt one of the strongest information instruments ever to turn out to be accessible to the general public, and professional se litigants who’re compelled to behave as each occasion and advocate concurrently ought to be capable of use it with the identical protections.

The court docket additionally made an essential statement in regards to the nature of AI itself. In contrast to a Google Doc or a typical piece of software program, AI carefully resembles “the type of confidential, strategy-laden iterative work product that Rule 26(b)(3) was designed to guard.” While you work together with an AI device, you’re sharing technique, testing arguments, and creating your case concept in a means that appears much more like speaking to a lawyer than placing one thing right into a Google Doc.

There’s a helpful analogy right here. Hundreds of attorneys and doubtless thousands and thousands of litigants use Gmail as their e mail server. We all know that Google trains on that data to a point, however we nonetheless prolong an inexpensive expectation of privateness to e mail. Choose Braswell’s dicta is questioning whether or not it’s time to increase that very same expectation to AI instruments, no less than for professional se litigants, and doubtlessly extra broadly sooner or later.

The brand new commonplace for AI and confidential data

That is the place the Morgan v. V2X Inc. opinion will get sensible. Choose Braswell’s amended protecting order units a transparent commonplace for when confidential data can be utilized with AI instruments.

The order learn, partly: “No occasion or approved recipient could enter, add, or submit CONFIDENTIAL Info into any fashionable synthetic intelligence platform, together with any generative, analytical, or massive language model-based device (“AI”), except the AI supplier is contractually prohibited from: (1) storing or utilizing inputs to coach or enhance its mannequin; and (2) disclosing inputs to any third occasion besides the place such disclosure is important to facilitating supply of the service…”

Anybody who has been following this area has most likely heard the saying, “In case you’re not paying for a product, you aren’t the client. You’re the product.” Choose Braswell is actually translating that precept right into a authorized commonplace. In case you’re utilizing the free model of ChatGPT, Gemini, Claude, or every other device, the data you present is probably going not stored confidential. A paid model with the precise settings could also be a special story.

The decide framed this as a pure extension of the unique protecting order. Don’t share confidential data with third events the place you may’t management its distribution, disclosure, or deletion. Now apply that very same logic to AI instruments, and look for a similar controls. And these obligations carry ahead by means of any AI device you utilize. In case you swap instruments mid-case, the identical commonplace applies to the brand new one.

There may be one factor the court docket might have accomplished higher right here. Within the professional se litigant abstract on the prime of the opinion, the decide wrote that “you might not add, enter, or submit Confidential Info into any mainstream AI device like commonplace ChatGPT, Claude, Gemini, or related platforms.” That abstract might go away the impression that professional se litigants can’t use these instruments in any respect. That’s not truly what the order says. It’s possible you’ll use ChatGPT, Claude, or Gemini in the event you can show that the correct safeguards are in place. A paid subscription with the precise settings and contractual phrases may meet this commonplace. We want the professional se abstract had made that clearer.

The entry to justice query

Choose Braswell acknowledged a stress in footnote 5 of the opinion. Legal professionals are utilizing subtle AI instruments with sturdy confidentiality protections, and spending actual cash to take action. Professional se litigants could not be capable of afford those self same instruments. The decide doesn’t resolve that battle, however she names it instantly: confidentiality protections shouldn’t be so costly that they’re out of attain for individuals who can’t afford a lawyer.

This issues as a result of the Morgan v. V2X Inc. case didn’t finish with the amended order. Simply final week, the defendant V2X filed a discover stating that the plaintiff’s disclosure of their AI use could also be inadequate. The plaintiff disclosed “Google” as their AI supplier and included Google’s commonplace phrases of service. The defendant argued that this isn’t particular sufficient. Google has Gemini, NotebookLM, and a number of other different AI merchandise, every with totally different capabilities and totally different information dealing with practices. The defendant needs to know which particular options the plaintiff is utilizing and which confidential paperwork are a part of that use.

That is the place a specificity requirement is rising. It will not be sufficient to call the seller. It’s possible you’ll have to show the way you’re utilizing the device and why your reliance on it carries an inexpensive expectation of confidentiality safety.

What Morgan v. V2X Inc. means on your agency

Morgan v. V2X Inc. offers attorneys a sensible framework for eager about AI and confidential data in discovery. Listed below are the takeaways.

Know what your AI device does along with your information. Learn the phrases and situations. Test the settings. In case your device or settings enable the device to coach on confidential information or shares confidential information unnecessarily with third events past the scope of offering its companies, it doesn’t meet the usual Choose Braswell set. A paid subscription with information safety ensures is a special product from a free public chatbot, and courts are actually drawing that distinction explicitly.
Be prepared to indicate your work. If opposing counsel challenges your use of AI, you want to have the ability to show that the device you used has contractual prohibitions in opposition to coaching on confidential information and disclosing them unnecessarily to 3rd events past the scope needed to offer the service. Doc which instruments you utilize, how you utilize them, and what settings you’ve got in place.
Identify the particular device, not simply the seller. The V2X defendant’s problem to the plaintiff’s disclosure tells us the place that is going. “Google” isn’t particular sufficient. “Gemini with information safety enabled beneath a paid Workspace account” is nearer to what courts will anticipate.
Anticipate extra of those fights. AI and discovery disputes are going to multiply. Till there’s a standard understanding of what these instruments can do and the way they deal with information, specificity of use will probably be litigated many times.

For attorneys who could discover professional se litigants’ use of AI irritating, it’s value stepping again and contemplating what Choose Braswell is actually saying. Individuals who couldn’t afford authorized recommendation now have entry to highly effective instruments that may assist them take part extra meaningfully within the justice system. That’s not an issue to resolve. That’s a growth value taking severely, and one which will finally make the system work higher for everybody.

Legal professionals who use purpose-built authorized AI instruments with applicable confidentiality controls are nicely positioned. The query isn’t whether or not you should utilize AI in litigation. It’s whether or not you may show that your device meets the usual. For corporations already utilizing instruments like Clio Work, that reply is sure.