As Minnesota’s cannabis market develops, we’re starting to see a pattern developing alongside it: cannabis industry members using AI tools to draft and generate legal documents. A primary reason for this is the belief that documents of sufficient legal quality can be drafted at a fraction of the cost of an attorney.
While in theory this makes perfect sense, this is absolutely not true in the cannabis industry. But in practice, we are seeing the downside of this approach increasingly:
-
Minnesota Cannabis Law is Not Generic, and AI Treats It Like It Is
Cannabis is one of the most legally complex industries as it operates within a patchwork of state regulations layered on top of continuing federal illegality. Most AI systems generate legal language by drawing from a large number of publicly available templates, blogs, and documents. These materials often come from many different states and industries, which are not fact-checked. As a result, the AI-generated agreements we see often include provisions that are irrelevant, legally incorrect, and even incompatible with cannabis laws in Minnesota. In heavily regulated industries, small drafting mistakes can create major compliance problems later.
Because the writing looks professional, cannabis companies assume the document is mostly correct. Minnesota cannabis law requires careful attention to regulatory detail. Treating these AI-generated documents like a standard business framework often leads to agreements that look complete on paper, but fail once regulators, investors, or disputes bring the underlying structure into scrutiny.
-
Saving Money Upfront Can Cost Leverage, Opportunities, and More Money Later
One of the most common reasons businesses use AI to draft legal documents is simple: it appears to save money. When a tool can generate a full agreement in seconds, paying an attorney for what appears to be the same work seems redundant. This seems like an easy choice to some cannabis applicants.
In reality, we often see the opposite outcomes.
When a client brings in an AI-generated agreement, the document usually looks polished, the formatting is clean, the sections are organized, and the language appears to be professional. But that polish makes the legal review process more complicated.
Because the document already looks complete, it requires a careful line-by-line review to determine whether the provisions are accurate, consistent, and compatible with Minnesota law and cannabis regulations. Most of the time, the attorney is not just spot editing but untangling existing language and even having to re-draft the entire contract. That process takes more time than drafting the document correctly from the start. Furthermore, AI-generated language can mask clients’ key objectives that an attorney must know to serve the clients well.
Beyond the cost of fixing the document, poorly structured agreements create real business risk. Unclear agreements create room to maneuver, and not always in the way that benefits the client’s business, which can lead to disputes, diminished leverage for negotiations, and so on. Legal documents are not just formalities; they define rights, responsibilities, and expectations. When language is ambiguous, it can weaken the very protections the agreement was supposed to create.
-
AI and Attorney Client Privilege
Even if someone were able to AI-generate a perfectly sound legal contract, there is another huge emerging risk that is becoming increasingly important: confidentiality.
Attorney-client privilege protects confidential communications between a lawyer and a client made for the purpose of seeking or providing legal advice. But that protection is limited. In a recent New York case[1], a court issued a bench ruling concluding that documents created using AI may not be protected by the attorney-client privilege because:
- Disclosure to a Third Party can Destroy the Attorney-Client Privilege Doctrine: AI functions as a third-party system, which is indicated in the AI’s company’s terms and services and privacy policy, and sharing sensitive information with AI platforms may be treated as disclosure to a third party, which can waive the privilege.
- AI is not an Attorney or Intermediary Limiting the Applicability of the Attorney- Client Privilege Doctrine: Attorney-client privilege only protects confidential communication between a client and a lawyer, and communication with other intermediaries such as translators. Here, AI is not considered an intermediary, so the attorney–client privilege does not extend to communications involving AI. The court would note that the documents were created independently, rather than for the purpose of obtaining legal advice or at an attorney’s direction, which may also result in a waiver of the work-product doctrine, since the materials were drafted solely at the client’s discretion.
- AI–Generated Documents Cannot Retroactively Become Confidential: The privilege must exist at the time of the communication, and thus, client-directed AI-generated documents cannot retroactively attach attorney-client privilege because the disclosure to a third party cannot cure the earlier waiver, even if the documents have been shared with the attorney.
In simpler terms, when business owners create AI-generated agreements, they may lose the legal protections that would normally apply. As a result, communications and documents that might otherwise remain confidential could become discoverable in legal proceedings, which can create significant legal risk and exposure.
Conclusion:
Minnesota’s cannabis market is still being built. Early decisions matter more than people realize. The documents clients create today will follow their business for years.
While AI can be great for certain aspects, it does not replace lawyers drafting legal documents. AI doesn’t know the stakes. It doesn’t carry liability. It won’t be there when regulators ask questions. The goal isn’t to draft something; it is to draft the right thing, in the right way, for the right regulatory reality. And that’s one place where shortcuts don’t pay off.
If you have questions about drafting legal documents and the benefits of an attorney consultation, contact Nicole Rash (nrash@hjlawfirm.com).
[1] See, United States v. Heppner, 25 CR. 503 (JSR), 2026 WL 436479, at *2–4 (S.D.N.Y. Feb. 17, 2026).
