Attorney Client Privilege In The AI Era
To read the article on Mondaq, click here
Recently, the world has seen landmark advancements in the field of artificial intelligence where one of the most popular innovation, has been the release of an application called Chat GPT, which has been described as an artificially intelligent chat-bot which generates user specific, human-like results for an individual. Chat GPT was designed by a company called Open AI, which is an AI research and deployment company based in the United States of America. Other than the said radical technology, the company was also responsible for the development of other artificially intelligent applications known as Dall E and Whisper, wherein the former works towards generating images from inputs of simple text, while the latter could be used to transcribe speech to text and for translating various languages.
Even though all three applications have attained significant popularity, it was only through Chat GPT that the company gained an un-phased momentum within all sectors, including the legal field, wherein, various legal professionals used the application to draft legal notices or contracts, for their clients. In September 2022, a new AI platform, named ‘Harvey’, which was described to be specifically designed for utilization by law firms, for their daily tasks. It was specified that the application, which was designed on a version of Chat GPT, used natural language processing, machine learning, and data analytics for automating and enhancing various aspects of legal work including contract analysis, due diligence, litigation and regulatory compliance.
Within their description, however, the law firm also specified that “Harvey’s output would need “careful review” by one of its lawyers – it comes with a disclaimer that it can still “hallucinate”, meaning it can produce inaccurate or misleading results”1, which gave a reasonable inference to the fact that there was a possibility that the outputs provided may be irrelevant to the query or even utterly deceptive. This reminds of the fact that the most technologically efficient machine that man has ever invented is the book. Indeed, the blindfolded slavery to such like softwares and platforms are only going to constrict human skill sets of logical reasoning, interpretation, perseverance which are much required for a seasoned lawyer.
Another question, which arose, from the usage of applications like Harvey (applications made specifically for use by Legal Professionals or ‘AILP’) or even from Chat GPT was a purported breach of the attorney client privilege, which is a generic name for an upheld doctrine in almost every country.
The Privy to Attorney – Client Privilege
The aforementioned term pertains to a general doctrine where the clients are offered with the privilege of keeping the conversation between them and their attorneys, private and confidential. In India, this doctrine is imbibed in the Indian Evidence Act, 1872, under Section 126 and Section 129.
Whereas Section 126 precludes any barrister, attorney, pleader or vakil (cumulatively referred to as ‘legal professionals’) to disclose any communications made to them, unless an express consent has been provided by the client, Section 129 precludes any individual from being compelled to disclose any confidential information between him and his legal professional adviser, unless he offers himself as a witness to the Court. The former provision, like the doctrine, provides certain restrictions, whereby, such communications can be disclosed, if the same is either made in furtherance of something illegal, or if the legal professional observes any crime or fraud being committed during the course of the employment. Even under clause 7 of the “Rules on an Advocate’s Duty Towards the Client” as published by the Bar Council of India, it has been clearly specified that the professional shall not, either directly or indirectly, disclose the communications made, either by the client to him or vice versa.
As per the laws of the United Kingdom, the privilege is classified under two distinct heads, vis a vis, legal advice privilege, referring to the communication between a client and his legal advisor, and litigation privilege, signifying various other communications including those between the legal advisor and potential witnesses.
In the recent case of Reliance Industries Limited vs. Securities & Exchange Board of India & Ors., the Hon’ble Supreme Court of India, while acknowledging and studying the privilege under various international laws, observed that,
“Indian position seems to be different from England. Section 126 to 129 of the Evidence Act do not draw any distinction between adversarial and investigative litigation as such, and privilege is applicable all through. This aspect is crucial, as it touches on the foundations of the legal profession at large in India.”
Under the laws of the United States of America, the privilege is covered by the Model Rules of Professional Conduct, according to which, while the client is entitled to confidentiality of communication, however, a third person who is privy to the communication between the attorney and the client is not entitled to the same.
A mere perusal of the aforementioned regulations and doctrines may, thus, give a reasonable inference to the fact that even if the structure may vary, the essence of privilege is well inducted in major laws of the world.
AI and the Privilege
It is needless to mention that, including India, the legal spectrum has yet to absolutely cover the application of various laws on artificially intelligent applications. In limine, it can also be specified that any information provided by a client or an attorney, in the case of AILP, may not be altogether between the client and the attorney and may even become liable to be disclosed before the Courts on account of the chatbot, not being classified within the definition of a legal professional.
A further danger to such confidential information is its disclosure to the developer or the company, which owns the chatbot and has a legal policy permitting the latter to use the volunteered information, for ‘research’ and other marketing purposes, which in the present case, is a policy of Open AI’s Chat GPT.
Not all Blues are Bad – The Positives of the AI
As specified earlier, an artificial intelligence app specifically designed to provide professional legal services pertaining to contract analysis, due diligence, litigation and regulatory compliance, may in fact make it possible for the legal professional to attain results without volunteering the name or details of the client or the communication that took place between the parties. It may also be possible for the legal professionals to only use such applications for obtaining information pertaining to the provisions or even the case laws in support of their case, which in turn would not require disbursing of any information privy between the professional and the client.
Thus, while such limited utilization of the application may limit the potential of the AI, it may however, release the AI as well as the professional, from a breach of doctrine, which in certain jurisdictions may invite a penalty and can further lead to disbarring of the professional.
This usage will further limit one of the most important and advertised function of such applications, which is legal advisory. It is pertinent to mention, that legal advice can only be offered after a certain kind and volume of personal information, relating to the facts and circumstances of the case, is volunteered by the client, which, in turn, will again start the loop of whether the AI, will be liable for breaching the privilege.
Conclusion – The Privilege – AI Conundrum
The developers and the primary users of AILP applications like Harvey periodically stress upon the fact that their applications have a potential to provide incorrect advise, and thus, may require strict overview of a lawyer. In order to add to its human quality, these applications may be entitled to the benefit of doubt when it comes to conducting errors in providing information.
Such benefit, however, may not be offered, in a case where the information volunteered to these applications, is furthered for usage, classified by the recipient of the information, as ‘research’ or for third party benefits. In fact, in cases where the application is utilized by a legal professional, the furthering of any information of this nature may even be contrary to the existing laws, rules, and regulations. This fact has also been recently realized by various countries, including Italy, which, on March 31, 2023, banned Chat GPT, after citing privacy concerns.
With the attorney-client privilege at stake, it may be time to either use the advancement of technology, in such form, where the information provided to the AI is not stored in its memory, thus reducing the chances of further disbursement. Another way of fixing this conundrum may be the limited usage of the application, as specified earlier. However, any such recommendation, without actually reviewing the functioning of the AILPs, and their potential usage of information may be considered to be presumptuous.
Therefore, it becomes essential for any legal professional or even a client to be aware of the kind of information that may be at stake, when the application is used for convenience.
- Victoria Basham, Allen & Overy integrates ChatGPT-style chatbot to boost legal work, Global Legal Post, Feb 16, 2023, available at: https://www.globallegalpost.com/news/allen-overy-integrates-chatgpt-style-chatbot-to-boost-legal-work-1735539269.
- Reliance Industries Limited vs. Securities & Exchange Board of India & Ors., AIR 2022 SC 3690.