Massachusetts is the newest jurisdiction to weigh in on the issues surrounding work product and attorney/client privilege for using AI chats. This case is a bit different from the others I have addressed. The difference here is that the plaintiff’s romantic partner is the one who asked ChatGPT for assistance.
The Facts
The case is Shealy v. Seaside Invs., LLC, No. 2684CV00799-BLS2 (Mass. Super. Ct. 2026). Shealy is a civil action involving loans and promissory notes. The defendant was seeking to compel emails between Shealy and his romantic partner, Fields, as well as documents addressing Shealy’s ability to pay the note. During discovery, “Plaintiffs redacted certain emails Shealy received from Fields which contained documents generated by use of ChatGPT.” Plaintiff emailed relevant information to Fields. She, in turn, on Shealy’s request, uploaded the documents to ChatGPT. ChatGPT generated a letter and documents, “which Fields emailed to Shealy.” In her explanation of her AI use, Fields claimed she was essentially “like a Xerox.” In response to defendant’s request for the letter and documents, plaintiff responded that “the ChatGPT documents were created in anticipation of litigation and qualify for work product protection.”
The Holding
The court found that “Plaintiffs’ argument is misplaced.” Certainly, litigation was anticipated and that is why the documents were created, but that is only part of the work product test. The other part of the test requires that the individual conducting the work meet the meaning of representative. The court listed the following as examples of a representative: “attorney, consultant, surety, indemnitor, insurer, or agent.” The list does not include romantic partner. Nor did plaintiff provide any cases supporting the notion that a romantic partner counts as a representative. Therefore, work product protection does not extend to the romantic partner of a party.
In its analysis, the court noted that “[t]he attorney work product doctrine ‘protects (1) documents and tangible things, [prepared] (2) by or for another party or by or for that other party’s representative…(3) in anticipation of litigation or for trial'”. The purpose of this doctrine is to “‘enhance the vitality of an adversary system of litigation by insulating counsel’s work from intrusions, inferences, or borrowings by other parties as [counsel] prepares for the contest.'” As a result, “[e]xpanding the definition of one’s representative to a romantic partner with absolutely no substantive connection to the litigation or to Plaintiffs’ counsel does not ‘enhance the vitality of an adversary system’ or insulate ‘counsel’s work from intrusions, inferences, or borrowings by other parties'”. As such, there is no basis to allow for such an expansion.
The next question the court analyzes is whether ChatGPT can be considered a ‘representative’. Here the court too finds that the answer is no. Most, but not all, of the previous opinions in which work product protection has been allowed are cases in which the party is pro se. As of this writing, only the Tate opinion has allowed for work product protection when the party is represented. In the pro se cases quoting Morgan, the instant opinion notes, “[t]he court appeared to conclude that the prompts submitted by pro se litigants to AI programs are akin to an attorney’s notes and thoughts about a case, and using the system does not ‘compromise work product protections.'” Shealy is quite different because the party in question is not pro se. Nor did plaintiff’s attorney ask Shealy to use ChatGPT. This, the court noted, makes Shealy more like Heppner, in which the court found that a represented party using AI of their own ‘volition’ is not entitled to protection. As in Heppner, it seems that the court in Shealy believes that the failure of counsel to ask plaintiff to use AI is fatal. I should note, however, that the recently decided Tate case disagrees with Heppner. In Tate the party used AI on its own volition, and not on the request of counsel, yet that court found work protect protection applied to certain AI chats
The end result is that the court chooses not to expand the work product doctrine to a case in which a party asks his romantic partner to use an AI tool. While I have not made it a secret that I find some of the jurisprudence related to privilege and work product protection as relates to AI quite concerning, I have no such concern here. I too see no basis to expand work product protection not only to one’s romantic partner, but to one’s romantic partner’s use of artificial intelligence.
If I were to find myself arguing whether a represented individual’s use of AI should be protected as work product, I would lean heavily on Tate while making my argument. Of course, opposing counsel would lean heavily on Heppner to argue the opposite.