Business Law

Using Generative AI in the Practice of Corporate Law

Generative AI (“GAI”), applied across a variety of industries, provides powerful tools to access large volumes of data, and its functions and capabilities have increasingly become more specialized in natural language processing, logical inferencing, machine learning, and image recognition technology. GAI refers to the capabilities of advanced technology to generate text, codes, videos, images, audio and other forms of content using various learning models. These models include deep learning, machine learning, statistical learning, large language models, and neural network models. Such models are able to learn patterns from large volumes of existing training data inputs to quickly create new content based on such inputs. GAI tools, such as ChatGPT, Microsoft’s Copilot, Westlaw’s Precision and Lexis+ AI, can analyze complex legal documents, quickly extract relevant information from case law, statutes and regulations, and even draft legal documents.

Many corporate attorneys and law firms are utilizing GAI tools developed specifically for the legal profession to supplement existing legal research tools, analyze contracts, improve templates, draft correspondence, conduct due diligence and streamline discovery. In M&A practice, GAI has the capabilities of sourcing target companies, conducting diligence more quickly and efficiently, and providing various analytic data to assist in the negotiation of deal terms. GAI holds the potential to streamline the legal process in evaluating a client’s legal issues and matters, reducing errors, and ensuring consistency.

When using GAI in the legal profession, attorneys should take into consideration privacy and confidentiality risks of client information, ensuring that the use of any GAI tools or materials do not jeopardize professional duties, including professional responsibility, diligence and attorney competency. The State Bar Standing Committee on Professional Responsibility and Conduct (COPRAC) has published its guiding principles on the use of GAI in the practice of law setting forth several duties of professional responsibility that should be considered when using GAI.[1]

With all practice areas within the legal profession, including corporate and transactional practices, accelerating the adoption of GAI, this bulletin looks at how GAI intersects with the practice of law, specifically corporate law, in California.

Privacy and Confidentiality

Client confidentiality is a critical aspect of developing the confidence between an attorney and client, and it provides the foundation for certain legal privileges including attorney-client privilege and the work-product doctrine. Clients seek such confidentiality to provide sensitive information to attorneys who are required to uphold a duty of confidentiality under the California Business and Professions Code (“BPC”) and under the California Model Rules of Professional Conduct (“RPC”).[2]

GAI tools, particularly ones that are available without a paid subscription, specify in the terms and conditions that any content provided in the tool will be used for purposes of developing, enhancing, and training the models. Because GAI tools use inputs that are provided when seeking prompts and learn from those stored prompts, such information is available to the developers of those tools and may be potentially revealed during future inquiries by third parties. To mitigate privacy risks related to the use of GAI, lawyers should use paid GAI tools that close out the data to only the firm and provide safeguards to restrict the work product which would not be accessible by any other third parties.

Particularly, California courts have continued to reinforce the protection of confidential information. In Sander v. State Bar of California,[3] the court looked into the safeguards to privacy and identification of admissions data in the plaintiff’s request to the State Bar of California to produce unidentifiable records for academic purposes. Under the BPC, identifying information submitted by applicants to the State Bar for admission and license purposes, including race, ethnicity, law school, transfer status, year of law school graduation, law school and undergraduate GPA, LSAT scores, and performance on the bar examination, are required to be confidential and shall not be disclosed pursuant to any state law.[4]

In Sander v. State Bar of California, the Court of Appeal examined the application of certain protocols including eliminating personal identifiable information from the admission data set and implementing additional steps to prevent re-identification. Plaintiff’s expert testified in trial court, arguing that the data could be made anonymous by a series of data manipulations. However, the Court of Appeal determined that such data manipulation would require the State Bar to “create a new record by changing the substantive content of an existing record or replacing existing data with new data” and determined that the State Bar of California would not need to produce such data as requested by the plaintiff, upholding confidentiality and privacy interests set forth in the BPC.[5]

In using GAI tools, care should be taken to safeguard client confidential information as required by the BPC and RPC. Attorneys should implement strict data protection measures and ensure that any GAI systems adhere to legal standards and guidelines set forth by the BPC and under the RPC. Licensed subscriptions of GAI tools tend to provide better privacy and security protections and limit the database to persons holding such valid licenses. In addition, attorneys should review the privacy policies, terms of use and other terms and conditions of GAI tools as it relates to privacy and confidentiality terms, how inputs are utilized, and the security protections of data inputs. Further, learning to set specific and precise prompts, while keeping client information confidential is important to reduce legal risks associated with the use of GAI tools.

Competency and Diligence

GAI has become increasingly popular in corporate practice, particularly in M&A transactions. However, the sophistication level and accuracy that is currently provided by GAI tools are still rapidly evolving, and most of the tools currently provided in the market are still working through accuracy and reliability issues.

Attorneys have a duty to competently and diligently provide legal services. As with any new technology, attorneys should continue to conduct reasonable diligence of any content that is provided by GAI tools. In general, using GAI tools as an initial draft, while maintaining client confidentiality can be a good starting point in preparing legal documents. As is routine in preparing drafts, attorneys should continue to monitor any GAI-generated outputs and verify them against current legal standards and precedents. In conducting research, GAI tools can provide another method of analysis to quickly produce results. However, such results should continue to be accompanied by supporting sources, fact-checking and editing.

Additionally, GAI tools present a known problem called hallucinations. Hallucinations occur when GAI tools provide false, inaccurate and untrue outputs, including improperly citing sources or citing fictitious sources.[6] GAI’s transformative capabilities to quickly access large volumes of data and provide responsive outputs raise concerns related to inferences and connections that are likely to result in inaccurate outputs.  Further, GAI tools do not always align with the nuances of legal terminology and requirements. When utilizing GAI platforms care should be taken to use certain customization options to tailor GAI outputs according to legal drafting styles, terminology preferences, while continuing to comply with the BPC, RPC and other legal standards. The capabilities of GAI do not replace legal analysis and critical decision-making processes in interpreting and resolving client issues that attorneys have been trained to do.

Compliance with Intellectual Property Law

Most open environment GAI tools provide terms that state that the developer of such tools are the owner of all, rights, title, and interest in and to the content that is generated by such tools. As such, keeping proprietary information of clients confidential is important and understanding that the content generated by the GAI tools do not belong to the user or the law firm. California lawyers have a duty to comply with the law.[7] This duty extends to the use of GAI tools in the practice of law and compels California lawyers to not counsel a client to engage in, or assist a client in, conduct that the lawyer knows is a violation of any law, including the law of intellectual property. Moreover, California lawyers should research and analyze the relevant intellectual property laws and concepts applicable to the attorney or the client in the matter when using GAI tools. Federal courts have held that only output authored or invented by a human being is protectable under current U.S. intellectual property law.[8] Further, the U.S. Copyright Office has indicated that whether any GAI tool’s output is protectable turns on, inter alia, whether the output was predictable in advance and on the specific prompts provided by the user of such GAI tool.[9]

A material issue in the use of GAI is whether the GAI tools infringe on intellectual property rights of the content owners, whose content was used to train and pull data for the outputs. Most open GAI review and analyze data from open online sources. Ongoing lawsuits, including the NYT v. OpenAI,[10] assert copyright and other intellectual property infringement as a result of the use of “unlicensed and unauthorized use and reproduction of … works during the training of its models.”[11] Derivative works developed from the outputs and contents created by GAI have the risk of copyright infringement.

To mitigate potential infringement risks associated with the use of GAI, attorneys should thoroughly review the terms, conditions and privacy policies of GAI tools and understand the intellectual property ownership of generated content. As previously mentioned, paid subscription models tend to provide a closed environment for such data to be kept private and not open to third party use.  

Training and Communication Regarding GAI Use

Workshops, MCLEs and other training programs on the use of GAI in the legal profession will help equip attorneys, paralegals, and other legal professionals with understanding GAI capabilities and its use. California lawyers have a duty to supervise lawyers and nonlawyers.[12] In the law firm setting, supervisory lawyers should set clear GAI use policies and make reasonable efforts to enforce these policies. At a minimum, these efforts include ethics training regarding use of GAI tools. No junior lawyer can be allowed to use GAI at the direction of a senior lawyer in a way that violates any lawyer’s professional responsibility, and this prohibition should comprise part of any training on the use of GAI tools in the legal profession.

California lawyers have a duty to communicate with their clients and a duty to reasonably consult with their clients regarding the means through which to pursue the objectives of the representation.[13] A lawyer’s communication obligations can evolve during the representation based on new technology, the multiplication of risks associated with GAI use, and client sophistication. If a lawyer intends to use GAI tools, then the lawyer should disclose to their client how the technology will be used, and the benefits and risks of such use. Client instructions and guidelines may need to be revised to the extent these may restrict or limit the use of GAI tools.

Creditworthy Claims and Candor to the Court

California lawyers have a duty of candor to the tribunal.[14] When GAI tools are used, this duty includes scrutiny of all GAI outputs, including analysis and citations, for accuracy before handing these outputs to a tribunal in any proceeding. In litigation, lawyers also have a duty to not take positions or assert claims or defenses that are unwarranted under existing law, unless a “good faith” extension, modification, or reversal of existing law can support the position. In each jurisdiction where a California lawyer practices, attorneys should also check for any rules or local orders requiring the disclosure of GAI use.

Last year, a New York federal judge sanctioned two attorneys who submitted a legal brief that included six fictitious case citations generated by an artificial intelligence chatbot, ChatGPT. In the sanctions order, the judge noted that the lawyers “continued to stand by the fake opinions” after the court questioned whether they existed.[15] Here in California, last October, a Los Angeles County Superior Court judge sanctioned an attorney who filed legal briefs citing to fake cases in unlawful detainer proceedings.[16] These fake cases appear to be common “hallucination” problems in programs like ChatGPT, which tend to produce things that look convincing, but actually have no basis in reality. Lawyers, whether working on litigated cases or on transactional deals, must watch out for these “hallucinations” and take care to not base positions or arguments or claims that are not creditworthy. 

Prohibition on Unlawful Discrimination

RPC prohibits California lawyers from unlawfully discriminating against persons on the basis of any protected characteristic when representing any client.[17] Some GAI is trained on biased information, and California lawyers need to apprise themselves of the risks created by using GAI tools that could result in unlawful discrimination (e.g., to screen potential clients or employees). Research conducted on biased data has been shown to result in “algorithmic discrimination.[18] For instance, facial recognition has been shown to be biased against black women, which Dr. Timnit Gebru and Dr. Joy Buolamwini showed in their Gender Shades study.[19] Another study, done by Ziad Obermeyer, showed an AI algorithm used to disperse medical funding had biases against black patients.[20] To avoid meeting the “knowingly permit” definition (i.e., to fail to advocate for corrective action where the lawyer knows of a discriminatory practice), attorneys should stay informed about GAI biases in legal practice and perform regular audits of GAI tools and systems that have produced biased results. Furthermore, law firm policies need to include ways to identify, report, and address potential GAI biases.

Legal Fees and Costs

California lawyers have a duty to not charge an unconscionable legal fee and to clearly state in a written contract the general nature of the legal services to be provided to the client, and any basis of compensation applicable to the matter.[21] Costs associated with GAI may be charged to clients in compliance with applicable law. This means that any fee agreement for legal services should explain the basis for the fees and costs associated with the use of GAI tools. Although lawyers can charge for actual time spent creating GAI inputs or editing GAI outputs, a lawyer cannot charge hourly fees for the time saved by using GAI tools.

Conclusion

GAI’s potential, like any other form of technology, if used appropriately, can increase productivity and accuracy, improve the quality of legal services and decrease legal expenses, among other things. However, integration of GAI tools in the actual practice of law requires consideration in the professional responsibility obligations discussed above. Thoughtful implementation and ongoing evaluation and auditing of GAI tools are essential to minimize the legal issues and concerns associated with GAI. As GAI evolves, attorneys should continue to stay informed on the proper uses of GAI in the practice of law.

This e-Bulletin was prepared by Shay Aaron Gilmore of The Law Office of Shay Aaron Gilmore and Durdana Karim of Stubbs Alderton & Markiles, LLP. Shay Aaron Gilmore and Durdana Karim are members of the Corporations Committee of the Business Law Section of the California Lawyers Association. The views expressed herein are those of the authors and do not necessarily reflect the views of The Law Office of Shay Aaron Gilmore or Stubbs Alderton & Markiles, LLP.


[1] State Bar of California, Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law, https://www.calbar.ca.gov/Portals/0/documents/ethics/Generative-AI-Practical-Guidance.pdf (Accessed July 7, 2024). See also, Suzanne Burke Spencer, Ethics Spotlight: New Guidelines for Lawyers Using Generative AI, California Lawyers Association, Jan. 31, 2024, https://calawyers.org/california-lawyers-association/ethics-guidelines-for-lawyers-using-generative-ai/.

[2]  See Bus. & Prof. Code, § 6068, subd. (e), Cal. Rules Prof. Conduct, Rules 1.6 and 1.8.2.

[3] Sander v. State Bar of California, 58 Cal.4th 300 (Cal. 2013).

[4] Bus. & Prof. Code, §6060.25, subd. (a).

[5] Sander v. State Bar of California, 58 Cal.4th 300 (Cal. 2013).

[6] Craig S. Smith, Hallucinations Could Blunt ChatGPT’s Success, Institute of Electrical and Electronics Engineers, March 13, 2023 https://spectrum.ieee.org/ai-hallucination.

[7] Bus. & Prof. Code, § 6068(a), and Cal. Rules Prof. Conduct, Rules 8.4 and 1.2.1.

[8] See, e.g., Thaler v. Hirshfeld, 558 F. Supp. 3d 238 (E.D. Va. 2021); See also 35 U.S.C. § 100.

[9] United States Copy Right Office, Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence, March 16, 2023, https://copyright.gov/ai/ai_policy_guidance.pdf.  

[10] See Rachel Reed, ChatNYT Harvard Law expert in technology and the law says the New York Times lawsuit against ChatGPT parent OpenAI is the first big test for AI in the copyright space, Harvard Law School, March 22, 2024, https://hls.harvard.edu/today/does-chatgpt-violate-new-york-times-copyrights/.

[11] See id.

[12] Cal. Rules Prof. Conduct, Rules 5.1, 5.2 and 5.3.

[13] Cal. Rules Prof. Conduct, Rules 1.4 and 1.2.

[14] Cal. Rules Prof. Conduct, Rules 3.1, 3.3 and 8.5.

[15] Sara Merken, New York lawyers sanctioned for using fake ChatGPT cases in legal brief, Reuters, June 26, 2023, https://www.reuters.com/legal/new-york-lawyers-sanctioned-using-fake-chatgpt-cases-legal-brief-2023-06-22/.

[16]David Wagner, This Prolific LA Eviction Law Firm Was Caught Faking Cases In Court. Did They Misuse AI?, LAist, Oct. 12, 2023, https://laist.com/news/housing-homelessness/dennis-block-chatgpt-artificial-intelligence-ai-eviction-court-los-angeles-lawyer-sanction-housing-tenant-landlord.

[17] Cal. Rules Prof. Conduct, Rule 8.4.1.

[18] See Joy Buolamwini and Timnit Gebru, Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification, Proceedings of Machine Learning Research, 81:1–15, 2018, https://proceedings.mlr.press/v81/buolamwini18a/buolamwini18a.pdf.

[19] See id.

[20] Linda Carroll, Widely-used healthcare algorithm racially biased, Reuters, Oct. 24, 2019, https://www.reuters.com/article/us-health-administration-bias/widely-used-healthcare-algorithm-racially-biased-idUSKBN1X32H8/.

[21] Cal. Rules Prof. Conduct, Rule 1.5 and Bus. & Prof. Code, §§ 6147–6148.


Forgot Password

Enter the email associated with you account. You will then receive a link in your inbox to reset your password.

Personal Information

Select Section(s)

CLA Membership is $99 and includes one section. Additional sections are $99 each.

Payment