A Massachusetts court has delivered another warning shot in the rapidly evolving debate over whether documents created with generative AI can receive traditional litigation protections.
In a recent ruling, a Massachusetts Superior Court judge rejected a claim that AI-generated documents created by a litigant’s girlfriend were protected under the work product doctrine. The court found that the materials did not qualify for protection because they were not prepared by an attorney or at the direction of counsel in anticipation of litigation.
The decision highlights a growing tension between how litigants increasingly use AI tools and how courts apply decades-old privilege doctrines. As individuals turn to ChatGPT and other AI systems to analyze disputes, prepare arguments, summarize facts, and organize evidence, courts are being forced to decide whether these AI-assisted workflows fit within existing legal frameworks.
The problem: AI has become a litigation assistant, but courts still see traditional roles
The work product doctrine generally protects materials prepared by or for attorneys in anticipation of litigation. The Massachusetts court focused on the fact that the documents were created by a third party, the litigant’s girlfriend, rather than by counsel or someone acting under counsel’s direction.
The ruling reflects a broader judicial trend: courts are increasingly treating consumer AI platforms as outside participants rather than extensions of the attorney-client relationship.
The issue is not simply whether AI was involved. The deeper question is whether the AI workflow preserves the legal relationship required for protection.
A document created by a paralegal following instructions from a lawyer may qualify. A document created independently by a spouse, friend, business partner, or AI chatbot may face a much higher hurdle.
The Massachusetts ruling follows a wave of AI privilege disputes
The case follows a series of recent decisions examining whether AI-generated materials receive attorney-client privilege or work product protection.
Earlier this year, a federal judge in the Southern District of New York ruled that documents created using Anthropic’s Claude AI platform were not protected by either attorney-client privilege or the work product doctrine. The case involved a defendant who used AI to analyze legal issues and prepare documents that were later shared with defense counsel.
The court’s reasoning centered on two issues:
First, communications with an AI system do not satisfy the traditional requirement that privileged communications occur between a client and attorney.
Second, the documents were created independently by the client rather than at the direction of counsel.
Legal commentators have warned that the growing use of consumer AI tools creates a potential privilege trap. A client may believe they are preparing material for their lawyer, but courts may view the AI platform as an outside third party.
The new AI litigation workflow problem
The legal industry is moving toward AI-assisted preparation at every stage of litigation.
Clients are using AI to:
- summarize documents before meetings with lawyers
- prepare timelines of events
- draft declarations
- analyze contracts
- organize evidence
- generate questions for depositions
- evaluate potential claims and defenses
The challenge is that many of these activities occur before a lawyer becomes directly involved.
That creates a gap between how people actually work and how privilege law defines protected legal work.
A client may view AI as a private legal assistant. Courts may view the same interaction as a conversation with an unrelated technology provider.
Enterprise AI platforms may become increasingly important
The rulings are likely to accelerate adoption of enterprise legal AI systems designed with stronger confidentiality controls, audit trails, and governance frameworks.
Law firms and corporate legal departments are increasingly evaluating AI systems based not only on accuracy and productivity, but also on whether their deployment supports legal obligations around confidentiality, privilege, and discovery.
The question moving forward will be less about whether lawyers use AI and more about whether AI usage is integrated into legally defensible workflows.
Implications for litigation teams
The emerging best practice for lawyers appears to be:
- Direct clients to use approved AI systems rather than consumer tools for litigation-related work.
- Document when AI tools are used under attorney supervision.
- Clearly establish that AI-assisted materials are created as part of counsel-directed legal work.
- Maintain records explaining confidentiality protections and intended purpose.
Courts are unlikely to ignore AI-generated materials simply because technology was involved. Instead, they are examining whether the human legal relationships surrounding those materials satisfy traditional privilege requirements.
AI is becoming evidence, strategy, and a discovery issue
The Massachusetts decision also reinforces a larger trend: AI systems themselves are becoming part of litigation.
Courts are increasingly confronting questions involving:
- AI-generated evidence
- AI-assisted legal research
- AI-created drafts
- AI chat histories
- discovery of prompts and outputs
- attorney obligations when using generative AI
A recent analysis of federal court opinions found that courts are generally applying existing legal doctrines to AI disputes rather than creating entirely new AI-specific rules.
For litigators, the message is becoming clear: AI does not exist outside traditional procedural rules. The technology may be new, but courts are applying old concepts of confidentiality, authorship, control, and legal purpose.
The next phase of legal AI adoption may depend less on whether lawyers can generate documents faster and more on whether those workflows can survive scrutiny in court.
Related AI and Legal Technology Stories
Federal court rules AI-generated legal documents are not privileged
A Southern District of New York ruling found that documents generated with Claude by a criminal defendant were not protected by attorney-client privilege or work product doctrine, even though they were later provided to defense counsel.
Law firms warn clients about AI privilege risks
Legal commentators have advised clients to avoid sharing sensitive case information with consumer AI systems because those conversations may not receive the same protections as attorney communications.
AI ethics issues continue to expand for lawyers
Lawyers are facing new ethical questions around confidentiality, accuracy, supervision, and responsible AI adoption as generative AI becomes embedded in legal workflows.
AI adoption is reshaping litigation practice
Legal professionals are increasingly using AI for research, document review, and litigation preparation, creating new opportunities alongside new procedural risks.
Sources
Minnesota Lawyer, “Massachusetts court rejects work product protection for AI documents by girlfriend”
https://minnlawyer.com/2026/08/03/massachusetts-court-rejects-work-product-protection-for-ai-documents-by-girlfriend/
Massachusetts Lawyers Weekly, “Work product privilege doesn’t protect AI docs created by girlfriend”
https://masslawyersweekly.com/2026/07/15/chatgpt-work-product-privilege-massachusetts/
Debevoise & Plimpton, “SDNY Rules AI-Generated Documents Are Not Protected by Privilege”
https://www.debevoise.com/insights/publications/2026/02/sdny-rules-ai-generated-documents-are-not-protect
Debevoise & Plimpton, “Update: Judge Rakoff Issues Written Opinion that AI-Generated Documents Are Not Protected by Privilege”
https://www.debevoise.com/insights/publications/2026/02/update-judge-rakoff-issues-written-opinion-that-ai
American Bar Association, “Federal Court Rejects Privilege Assertion for AI-Generated Research”
https://www.americanbar.org/groups/litigation/resources/newsletters/intellectual-property/federal-court-rejects-privilege-assertion-for-ai-generated-research/
Nixon Peabody, “AI-generated documents may not be privileged: Key takeaways”
https://www.nixonpeabody.com/insights/alerts/2026/02/25/ai-generated-documents-may-not-be-privileged