Risks of Using AI for Legal Advice
Artificial intelligence (AI) tools like ChatGPT are revolutionizing how individuals and businesses access information, draft content, and explore solutions. As transactional attorneys, we see immense potential in AI, but also significant pitfalls when clients use these tools for legal research, contract drafting, and other sensitive legal matters without professional oversight.
Here are five common risks to be aware of when turning to AI for legal guidance:
1. Inaccurate or Outdated Information
AI models are only as good as the data they were trained on, and they do not update themselves in real-time. Relying on AI for legal research may yield results that are outdated or simply incorrect, especially in areas where the law evolves quickly.
We recently received an AI generated inquiry from a client claiming there were additional securities filing requirements associated with a transaction. After looking into this claim, we discovered that the AI model was pulling its information from a proposed change to the law from over ten years ago, along with various legal papers on the effects of such proposal. However, such proposal was never enacted into law. Solely relying on the confident responses of AI models can be detrimental, as these outputs are generated based on patterns in data and should not be treated as definitive or authoritative sources. Solely relying on the confident responses of AI models can be detrimental, as these outputs are generated based on patterns in data and should not be treated as definitive or authoritative sources. Particularly when dealing with factual matters, AI-generated statements are no substitute for the independent verification and thorough research that an attorney can provide. Notably, AI models can “hallucinate” statutes or case law that do not exist. Clients should always verify key authorities by citing primary sources such as Westlaw or Lexis.
2. Lack of Legal Nuance
Unlike legal professionals, AI tools lack the ability to weigh strategic implications, jurisdictional subtleties, and the business contexts that shape contract language and transaction structuring. A clause that appears acceptable in a vacuum may be commercially impractical or legally unenforceable in a client’s specific situation.
We recently received comments on a contract from a client who had run the agreement through an AI tool. The comments addressed surface-level grammatical and formatting edits but did not consider the substance of the contract, analyze how the terms aligned with the client’s business objectives, or identify terms that were not “market” for this type of transaction. Without an attorney’s legal review, the client could have missed critical negotiation points to raise with the other party. Moreover, AI tools are not subject to the ABA Model Rules of Professional Conduct. They cannot perform conflict checks, safeguard privilege, or fulfill fiduciary duties in the manner a licensed attorney must.
3. One-Size-Fits-All Solutions
AI often produces generic contracts and templates that may omit critical deal-specific terms or fail to address industry-specific regulations. We have seen clients use boilerplate documents that did not include essential protections related to intellectual property rights, indemnification, or dispute resolution, leading to expensive clean-up work or even litigation.
Issues also arise when unique clauses or heavily negotiated terms are misunderstood or mischaracterized by AI tools, especially if those terms do not match standard patterns the AI has seen in other examples. As with legal nuance, these generic forms are created without considering the practical implications of certain clauses or omissions, such as enforcement risks, market norms, or client-specific goals. AI tools may misinterpret custom terms, resulting in omissions or mischaracterizations. Unlike attorneys, they do not maintain version control logs or audit trails, both essential in litigation and regulatory review.
4. No Liability or Accountability
If a client relies on inaccurate or insufficient legal advice from an AI model, there is no malpractice claim to file and no firm to hold accountable. When a client engages a law firm, it benefits from licensed professionals who stand behind their advice and are ethically bound to protect client interests.
As market conditions and deal terms fluctuate, attorneys can have ongoing conversations with clients to address those changes and make informed adjustments. If any deal terms are later disputed, written communications between attorneys and opposing counsel can be used to defend positions in court and support arguments that protect client interests. Courts have not been sympathetic to parties that rely solely on AI, particularly when they fail to seek professional assistance to protect their rights.
5. Confidentiality Risks
When sensitive information is inputted into public AI tools, clients may be unknowingly exposing proprietary data. Although AI platforms are continually improving their privacy protocols, they are not a substitute for the attorney-client privilege and the secure systems used by law firms to protect confidential information.
Without a clear understanding of how an AI tool stores, processes, or reuses data, confidential information may be retained or inadvertently disclosed. This could reveal trade secrets, breach non-disclosure and confidentiality agreements, and cause significant harm to the client, their business, and their relationships.
Under ABA Model Rule 1.6, attorneys are bound to protect client confidentiality. Most AI platforms, by contrast, include terms of service that disclaim any duty of confidentiality and reserve the rights to store, process, and reuse input data for training purposes.
What AI Can Be Good For
While AI should never fully replace professional legal counsel, there are plenty of very useful ways these tools can be used by clients to save time, money, and valuable resources:
- Initial brainstorming for business transactions, negotiation discussions, and meeting preparation. Assigning an AI tool a role, detailing a situation, and asking it to simulate meetings or discussions can help prepare clients for potential questions and issues that may arise.
- Assisting with initial drafts of memos, contract provisions, or communications, providing a base for clients to refine and finalize.
- Understanding basic legal terminology and concepts prior to meeting with an attorney, helping bridge the gap between legal language and business goals.
- Summarizing documents or statutes for internal discussions or to share with non-legal team members.
The key takeaway: AI is a tool, not a complete legal strategy. It may have immense computing power, but it cannot replace Attorney Insight.
At FBFK Law, we are proud to combine technological savvy with decades of legal experience to serve our clients better. We encourage clients to bring their AI-generated questions or drafts to us, so we can vet, refine, and tailor them to fit the unique legal and business goals of our clients.
