Regulating AI In Legal Services - Professional Ethics And The Question Of Liability
Author: LawSphereX
Category: Technology, Cyber & Privacy
Published:
Updated:
REGULATING AI IN LEGAL SERVICES: PROFESSIONAL
ETHICS AND THE QUESTION OF LIABILITY
Vera Kotian, Kirit P. Mehta School of Law (NMIMS)
ABSTRACT
Artificial intelligence has rapidly moved from a futuristic concept to an everyday tool in legal practice, promising unparalleled efficiency in research, drafting, compliance, and litigation support. However, the accelerated integration of AI raises serious ethical concerns that existing professional frameworks in India are unprepared to address. This paper explores how AIinduced errors-hallucinations, fabricated citations, and misinterpretationschallenge fundamental obligations imposed on advocates by the Bar Council of India Rules. The paper illustrates, through incidents such as Mata v. Avianca and the Buckeye Trust v. PCIT order, how blind reliance on unverified AI could mislead courts and undermine judicial reasoning.
It further considers the growing risk to client confidentiality with the increasing use of legal professionals uploading sensitive information to AI systems that have inadequate privacy protection and may retain data indefinitely. The discussion then shifts to the evolving question of liability, considering developments such as the Product Liability Directive of the EU with a view to determining whether liability can extend beyond lawyers to developers and other actors in the AI supply chain. Thereafter, the paper argues that, irrespective of progress in technology, the ultimate duty of oversight, accuracy, and ethical compliance remains with legal professionals.
Finally, it concludes by emphasising the need for clearer regulatory guidelines, training in AI literacy, and more stringent transparency requirements in ensuring that AI supports, rather than compromises, the values underlying the legal profession. Keywords: artificial intelligence, legal ethics, AI hallucinations, client confidentiality, professional liability
INTRODUCTION
“Each new technology comes with its own set of ethical pitfalls and, unfortunately, cautionary tales from lawyers who fail to mind them”1 In the span of three years, AI has evolved from a futuristic concept found in science fiction books to being an inseparable part of our day-to-day reality. The permeability of AI in every field and profession is immeasurable, and the legal field is no exception to that. So, what happens when the same AI used in legal discourse errs? Who is held liable for the mistakes of a machine, and where do the standards of professional ethics apply in such scenarios? AI assists in a wide range of tasks, including document review, legal research, document summarisation, deposition questions, and helps ensure contract policy compliance.
2 When technical and more administrative work is given to AI, it helps reduce the burden on lawyers and instead helps them to focus more on the substance and outcome of cases. With great power comes an even greater ethical responsibility. Unlike traditional machines or tools, which are entirely controlled by human operators, many AI systems function autonomously and make decisions based on algorithms that evolve over time. This autonomy creates challenges for the application of existing legal principles, which typically rely on human intent, control, and predictability3 Because of its autonomy, AI cannot be just a neutral assistant or a passive tool. Biased results, hallucinations, or mistakes can have detrimental effects, ranging from deceiving judges to endangering client privacy.
AI is not human and therefore cannot be asked about its reasoning directly, which is why cautious verification and oversight are of utmost importance. This paper seeks to understand the challenges and offer perspectives on where we stand as legal professionals in a field increasingly defined by AI. 1 Cassandra R. Hewlings, Future of Louisiana’s Ethics and Professionalism Rules: As Technology Changes, Will Ethics Stay the Same? , LOUISIANA BAR J. , June/July 2016, at 42, 42. 2 Marjorie Richter, See what legal professionals say about the role of AI and law, Thomson Reuters Law Blog (Aug. 18, 2025), https://legal. thomsonreuters. com/blog/how-ai-is-transforming-the-legal-profession/. 3 Shreya Goyal, AI And Legal Liability: Who Is Responsible When Machines Go Wrong? , 8 Indian J. L. & Legal Rsch. ,(2025).
WHAT IS GENERATIVE AI?
AI has been defined as the science of teaching computers how to learn, reason, perceive, infer, communicate, and make decisions as humans do. 4Generative artificial intelligence (GAI) refers to a group of AI algorithms and models that are capable of producing new content, including texts, images, videos and problem-solving strategies, with human-like creativity and adaptability5. Popular examples of these generative AIs would be ChatGPT, Claude, Gemini, Perplexity, DALL·E, and many more. While these seem to be only recent developments, the history of Gen AI actually traces back all the way to the 1950s, with the advent of rule-based systems. A rule-based system operates through a sequence of steps. It first receives input, like numbers, facts, or even symptoms. It then matches this input with a series of "if–then" rules embedded in its knowledge base.
If a rule is found to match, the system executes it and performs the associated action. When two or more rules are applicable simultaneously, the system applies techniques to determine which one to execute first, typically by selecting the most specific or highest-priority rule. Last but not least, the system produces an output, which may be a decision, recommendation, or reply, for example the General Problem Solver (GPS), developed in the 1950s by Herbert A. Simon and Allen Newell designed to solve problems like humans by breaking down problems into smaller sub-problems. We also have MYCIN, an expert system developed in the 1970s to diagnose bacterial infections and recommend antibiotics6. This soon evolved into today’s generative AI, which is a novel AI technology that can produce new content automatically by utilising input data. The theoretical framework of generative AI comprises
1. Machine learning: an application of Artificial Intelligence (AI) that provides systems
the ability to automatically learn and improve from experience without being explicitly programmed. It is a continuous process in which a program or a system learns and improves from its experience. It makes an intelligent prediction of an outcome based
4 John Simons, Tomorrow’s Business Leaders Learn How to Work with A.I., WALL STREET JOURNAL,
November 30, 2016, available at https://www.wsj.com/articles/tomorrows-business-leaders-learn-how-to-workwith-a-i1480517287 5 Ran He, Jie Cao & Tieniu Tan, Generative artificial intelligence: a historical perspective, 12 Nat'l Sci. Rev. 1, (2025), https://academic.oup.com/nsr/article/12/5/nwaf050/8029900. 6 Jeff Johnston, The History of AI: From Rules-based Algorithms to Generative Models, Lantern (July 17, 2024), https://lanternstudios.com/insights/blog/the-history-of-ai-from-rules-based-algorithms-to-generativemodels/.
on data dataset.7
2. Natural language processing (NLP): a multidisciplinary field that focuses on enabling
machines to understand, interpret, and generate human language in a way that is both meaningful and contextually relevant. It combines principles from linguistics, computer science, artificial intelligence, and cognitive psychology to bridge the gap between human communication and machine understanding.8
3. Image processing, and computer vision9.
Current use of Gen AI in the legal profession The vast capabilities of these models, especially large language-based models like ChatGPT, make them extremely attractive to those in the legal profession. These models have the ability to compute large amounts of information at a fraction of the time, which makes routine but intensive work like drafting of contracts, basic legal research and document review much more scalable and cost-effective. Even though the technology is not at a stage where lawyers can be replaced10Through machine learning and natural language processing, the reasoning and output of the AI are similar to those of a human being. Multiple types of “legal” AI have emerged in the span of a few short years. Some of the wellknown GenAI products currently being deployed in the legal sector include Microsoft Copilot, Lexis+ AI, Thomson Reuters Cocounsel and Harvey.
11 These have specifically been made to act as assistants to lawyers. They help overcome natural human errors that most lawyers can make when trying to understand large amounts of information. Certain tools utilise NLP to analyse court dockets, predicting decision outcomes, identifying favourable jurisdictions in which to file a lawsuit, and proposing likely successful motions and arguments before specific
7 Alam Azmir, What is Machine Learning?, 2023, ResearchGate, https://doi.org/10.5281/zenodo.8231581.
8 Rayhan, A., Kinzler, R., & Rayhan, R. (2023). NATURAL LANGUAGE PROCESSING: TRANSFORMING
HOW MACHINES UNDERSTAND HUMAN LANGUAGE. ResearchGate. https://doi.org/10.13140/RG.2.2.34900.99200 9 Zhihan Lv, Generative artificial intelligence in the metaverse era,Cognitive Robotics,Volume 3,2023,Pages 208-217,ISSN 2667-2413,https://doi.org/10.1016/j.cogr.2023.06.001. 10 Steve Lohr, A.I. Is Doing Legal Work. But It Won’t Replace Lawyers, Yet., NEW YORK TIMES, March 19, 2017, available at https://www.nytimes.com/2017/03/19/technology/lawyers-artificial-intelligence.html 11 Alex Heshmaty, The State of AI and Legal, Legal Management (Oct. 11, 2024), https://www.alanet.org/legalmanagement/2024/october/features/the-state-of-ai-and-legal.
judges.12 AI is currently being deployed to do the following work
1. Legal research: by instantly scanning the Internet for legal principles, case laws, and
doctrines, AI reduces the manual workload most legal professionals are faced with.
2. Contract drafting and review: with AI, one only has to keep a well-worded prompt and
a contract can be generated in seconds. It can also review and flag clauses in the contract that are vague, unnecessary or wrong.
3. Litigation support: AI can imitate human reasoning and think strategically, with this it
can predict case outcomes, judgment patterns and thus support an advocate when their making their case.
4. Compliance: AI can review policies and regulatory frameworks and find potential
loopholes and inconsistencies. According to a report from LexisNexis, around a quarter of lawyers regularly use GenAI tools, compared to just over 10% a year ago. Still, almost 40% of legal professionals surveyed said they had no plans to adopt GenAI.13 This data displays a profession in flux, with a divide between those who view AI as a welcome addition to the legal profession and those who see the probability of misutilization of this technology in the field. Addressing the ethical and legal ambiguities surrounding AI is essential to bridging this gap. How do these principles change in practice? What obligations do lawyers have when using AI, and most importantly, who is accountable when AI fails?
THE STANDARDS OF PROFESSIONAL ETHICS TO AI-INDUCED ERRORS IN
LEGAL PRACTICE
The body of Advocates, which determines the standard of conduct for the members, is the Bar
12 Anirudh Gotety, REGULATING THE ETHICS OF THE UNKNOWN: ANALYSING REGULATORY REGIMES
FOR AI-BASED LEGAL TECHNOLOGY AND RECOMMENDATIONS FOR ITS REGULATION IN INDIA,
14 NUJS L. Rev. 3 (2021).
13 Laura Halls, Exploring the growing appetite for legal AI | LexisNexis Blogs, LexisNexis | Legal and Professional Solutions and Products (Mar. 27, 2024), https://www.lexisnexis.co.uk/blog/future-of-law/explorethe-growing-appetite-for-legalai#:~:text=The%20report%20shows%20adoption%20rates,trend%20looks%20likely%20to%20continue.
Council of India. The Standards of Professional Conduct and Etiquette are part of Chapter II of the Bar Council of India Rules.14 The Bar Council of India (BCI), constituted under Section
4 of the Advocates Act, is empowered under Section 49 to prescribe standards of professional
conduct and etiquette for advocates15.These rules, however, have not kept pace with the evolving legal practice in the country. The BCI Rules were imagined in reference to and in context of a world where a lawyer practices law only in a courtroom.16 Moreover, it imagines a world where technology is kept away from the courtroom. These rules are silent on issues such as technological competence, AI oversight, or client consent in AI-assisted legal services17This silence allows for gaps and breaches in the level of professionalism a lawyer must uphold.
I.MISLEADING THE COURT: AI HALLUCINATIONS
Artificial Intelligence hallucinations refer to instances where AI systems generate outputs that are factually incorrect, misleading, or fabricated, often with a convincing degree of plausibility18 It has been known to a. Reference the correct legal principle while citing non-existent cases, with seemingly correct methods of citation b. Fabricate legal doctrines and principles which either do not exist or cannot apply to a certain jurisdiction c. Misinterpret or incorrectly summarise case laws d. Arrive at false legal conclusions or apply repealed or obsolete laws 14 Neeraj Sharma, Ethical Lawyering: A Way to Uphold the Nobility of the Legal Profession, 4 Int'l J. L. Mgmt. & Humans. 488, (2021), https://doij. org/10. 10000/IJLMH. 111359. 15 3 Bar Council of India Rules, Part VI, Chapter II, "Standards of Professional Conduct and Etiquette," available at: http://www. barcouncilofindia.
org 16Gotety, supra note 11 17 Sreddha Gopakumar Krishna, Ethical And Professional Responsibility In AI- Enhanced Legal Practice: Need For A New Code Of Conduct, 7 Indian J. L. & Legal Rsch. , (2025).
18 Debajyoti Chakravarty, AI Hallucinations in the Legal Field: Present Experiences, Future
Considerations, OBSRVER RESEARCH FOUNDATION (Aug. 12, 2025), https://www.orfonline.org/expertspeak/ai-hallucinations-in-the-legal-field-present-experiences-future-considerations.
When used as a quick solution, it can lead to disastrous consequences. In Mata v. Avianca19 The plaintiffs' attorney relied on ChatGPT to find cases that turned out to be false. These were submitted in a court filing when the court noticed that the cases were false, leading to sanctions on the attorney. This underscores that AI-like ChatGPT are not legal databases and therefore can hallucinate. It is the ethical responsibility of thelawyer to defend and represent the client fearlessly and with fair and honest means, regardless of their personal opinion. (Rule 15 BCI rules). 20 In a similar case of Buckeye Trust v. PCIT21 The Bengaluru bench of the Income Tax Appellate Tribunal (ITAT) issued an order referencing three Supreme Court judgments and one Madras High Court ruling, all of which were generated by AI and non-existent.
Within a week, the order was swiftly withdrawn, citing “inadvertent errors”. 22 This case highlights a much deeperseated issue. Till now, AI has been used by lawyers to enhance legal research, which implies that the ethical responsibilities and the consequences of a breach will fall on the lawyers themselves, but in this case, the AI was used as part of judicial reasoning by a certified tribunal, the tribunal had to use the reasoning and submissions of both parties, which makes it unclear how much of the reasoning could be influenced by AI. This also makes it harder to contest or appeal. This also shows that using AI ethically falls on all participants of the judicial process. Necessary recommendations Preventing the risks of AI hallucinations requires a multi-layered approach combining policy frameworks, technical safeguards, professional responsibility, and judicial oversight.
India's NITI Aayog’s Responsible AI – Approach Document, Part 1 (2021)23 sets down guidelines for AI in India and emphasises safety, accountability, transparency and reliability. There must be constant error testing of the AI, documentation and external audits conducted.
19 Mata v. Avianca, Inc., U.S.D.S.D. OF N.Y., June 22, 2023,
2022cv01461, (U.S.), https://law.justia.com/cases/federal/district-courts/newyork/nysdce/1:2022cv01461/575368/54/. 20 Bar Council of India Rules, Part VI, Chapter II, Section II, Rule 15, Standards of Professional Conduct and Etiquette (as amended), available at https://www.barcouncilofindia.org/info/rules-on-an-advocates-dutytowards-the-client
21 ITA No.1051/Bang/2024 Buckeye Trust, Bangalore
22 counselvise, AI Hallucination in Court: ITAT Recalls Order Over Fake ChatGPT-Generated Case Laws in Buckeye Trust Tax Dispute, Counselvise - AI-Driven Legal Research & Expert Consultations (Mar. 20, 2025), https://counselvise.com/blogs/ai-hallucination-itat-buckeye-trust. 23 NITI AAYOG, India | Approach Document for India Part 1- Principles for Responsible AI, NITI AAYOG, India (Feb. 20, 2021), https://www.niti.gov.in/node/326.
The developers of AI systems must prioritise the quality of data over the quantity. This is done by training the AI in current, diverse and correctly sourced information. It should be mandatory to have an in-built transparent design feature from the very start, which will disclose sources, model architecture and performance benchmarks. Users should be able to trace the sources of the material provided by AI, which means ensuring verified citations for all the information. It should be ensured that contracts with AI service providers include provisions that require transparency regarding the AI system's training data and algorithms. 24 Furthermore, courts must evolve and adopt newer rules and regulations to incorporate AI. It must be disclosed where and how AI has been used in a legal proceeding.
Bar councils can issue binding guidelines to ensure human verification of AI-generated legal content. In order to prepare attorneys to cross-check and report mistakes, law firms and judicial institutions could also create internal governance codes that emphasise AI literacy. A strong framework of ethical standards, technical transparency, and human oversight is
necessary to fully utilise AI's legal benefits without sacrificing integrity. AI can only be used as a tool to promote justice rather than skew it if developers, courts, regulators, and practitioners work together with vigilance.
II. CLIENT CONFIDENTIALITY AND DATA SECURITY
“OpenAI CEO Sam Altman has warned that while users often reveal the most personal details of their lives to ChatGPT, their interactions lack privacy protections and could potentially be produced for lawsuits or other legal reasons.”25 this one line from a Hindu article is enough to underscore one of the biggest ethical problems in integrating AI with the legal field. AI works on a system of giving information to get information. When legal professionals need AI review policies or to draft contracts, it involves the process of uploading large amounts of data, most of which could be sensitive. 24 Kiara Fink, AI Hallucinations In Legal Practice: Risks, Real Cases, And Solutions, Mondaq - Law Articles and Insights (Nov. 7, 2024), https://www.mondaq.com/new-technology/1540712/ai-hallucinations-in-legalpractice-risks-real-cases-and-solutions.
25 OpenAI CEO Sam Altman warns ChatGPT users’ personal questions could be used in lawsuits, The
Hindu (July 25, 2025), https://www.thehindu.com/sci-tech/technology/openai-ceo-sam-altman-warns-chatgptusers-personal-questions-could-be-used-in-lawsuits/article69864881.ece.
Attorney-client privilege is intended to facilitate candid and comprehensive communication, thereby enabling clients to seek legal guidance without apprehension that their disclosures might be exploited against their interests. 26Clause 7 of the "Rules on an Advocate's Duty Towards the Client"27 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. Furthermore, section 134 of the Bhartiya Sakshya Adhiniyam upholds attorney-client privilege, preventing lawyers and clients from disclosing confidential information exchanged during legal employment, except in cases of ongoing crimes or fraud28.
When sensitive data is uploaded, lawyers lose effective control over where the information is stored, how long it is retained for and who can access it. Since AI gets trained on the data fed to it, there's a strong possibility that it uses that data for future outputs, causing leaks. A lawyer's duty of confidentiality is not a passive one; the very act of using an AI tool and uploading data onto it without reviewing its privacy policies is an act of negligence. In the recent order of OpenAI, Inc. Copyright Infringement Litigation29a federal judge ordered OpenAI to preserve every ChatGPT conversation going forward, including those users try to delete, as part of an ongoing copyright lawsuit brought by several news organisations.
30 This case shows us that any data one uploads to AI, whether personal or professional, can be preserved for an indefinite amount of time and be used in litigations completely unrelated to certain parties. India's laws on evidence and crime have already been updated to include all forms of electronic evidence, and while it's not explicitly in motion right now, AI-based records will also be accepted as evidence in most courts in a few years. While an advocate must do everything in his power to ensure confidentiality, it should also be a state's responsibility to introduce new laws and mechanisms to ensure privacy from the get- 26 Ismail Amin, Attorney-Client Privilege at Risk with AI Platforms, TALG®, https://talglaw. com/artificialintelligence-platforms-and-the-potential-waiver/ (last visited Sep. 20, 2025).
27 Bar Council of India Rules, Part VI, ch. II, § II, cl. 7 (under the Advocates Act, 1961).
28 Bharatiya Sakshya Adhiniyam, No. 46 of 2023, § 134 (India).
29 In Re: OpenAI, Inc. Copyright Infringement Litigation, 1:25-md-03143, (S.D.N.Y.)
30 Kyle Moen, Think your AI Chats are Private? The Court Might Disagree. | Segal Duffek Moen | Minneapolis, MN, Segal Duffek Moen (June 19, 2025), https://sdmlawyers.com/think-your-ai-chats-are-private-the-courtmight-disagree/.
go. For example, the Tamil Nadu e-Governance Agency (TNeGA) implemented the DEEP- MAX Scorecard to evaluate AI systems before public rollout. This scorecard ensures AI systems align with ethical standards by assessing diversity, equity, and misuse protection. The scorecard covers various aspects, including data quality, algorithmic fairness, and impact assessment.31 A lawyer's duty of competence and upholding the lamp of industry
The ethical dilemma of using AI for efficient work vs upholding client confidentiality will remain in contention for years. We must not forget a lawyer's duty to be competent and industrious while also evolving with society's changes. The duty of competence is not restricted to mastery over the procedural and substantive laws anymore; it involves being fully able to understand the entire legal system and the tools that shape it. With the growth of AI, it is inevitable that legal professionals have to make provisions to include it in their day-to-day instead of actively avoiding it, and that starts where the lawyer is competent in the laws of AI. While the duty of competence is not directly enshrined in the BCI rules, it is implied in the existing duty to the client. It is, however, explicitly mentioned in Rule 1. 1 of the American Bar Association (“ABA”) rules32 Principle 3 and Rule 3.
2 of the Solicitors Regulation Authority (SRA) Code of Conduct (2023)33, Rule 5(2)(c) Legal Profession (Professional Conduct) Rules, Singapore34. The Lamp of Industry signifies hard work. An advocate should give their client’s case their all. He should read literature, know relevant developments in the law, and organise his thoughts for stronger arguments. The attorney should give the case everything they have. 35 Attorneys are expected to stay abreast of technological developments that can enhance their work36 and must adequately supervise the use of AI tools by their staff and ensure the accuracy of AI-generated work37To implement these duties, legal professionals must be trained in how to efficiently and responsibly use AI. 31 Ispp, AI Ethics in Public Policy: Case Studies & Challenges | ISPP, Indian School of Public Policy - (July 10, 2024), https://www. ispp. org.
in/ethics-of-ai-in-public-policy-in-the-indian-context/.
32 Model Rules of Prof’l Conduct r. 1.1 (Am. Bar Ass’n),
33 SRA Code of Conduct for Solicitors, RELs, and RFLs (2023), Principle 3 and Rule 3.2
34 Legal Profession (Professional Conduct) Rules 2015, Rule 5(2)(c), Sing. Subsidiary Legislation No. S
35 Ruchit Yadav, PROFESSIONAL CONDUCT AND LAMPS IN ADVOCACY - Jus Corpus, Jus
Corpus (Aug. 16, 2023), https://www.juscorpus.com/professional-conduct-and-lamps-in-advocacy/. 36 Terzidou, K. Generative AI systems in legal practice offering quality legal services while upholding legal ethics. International Journal of Law in Context 2025, pp. 1–22.
37 Generative AI for Legal Professionals - Promise, Peril, and a Path Forward, 2023.
At the very base, AI training should be taught at the university level, it should be mandatorily included in the curriculum, emphasising on skills such as identifying the correct AI for a particular task, prompt engineering, and how to maintain ethical standards when dealing with AI. Next, there should be mandatory training of lawyers in law firms. on getting equipped with AI. They should be trained in how to construct the right queries and evaluate the relevance, quality, and accuracy of the responses (and then update the queries as needed), and be able to synthesise the overall results into a cohesive, actionable picture38. Internationally, many law firms have implemented summer programs to teach their employees how to effectively use AI.
Summer associates learn how to use the research and chatbot tools used by the firm for tasks like legal writing, document creation and summarisation, drafting, proofreading, and more. This development signals a commitment to new technology, paving the way for young lawyers to develop these critically important competencies. 39The training given should be hands-on and human-to-human; it should demonstrate tools in context and overall create a community built on AI practices.
III. BEARING THE LIABILITY WHEN AI ERRS
One of the most relevant questions asked in the topic of AI and its integration with the legal field is who bears liability when an AI makes a mistake. While on the face of it, an advocate’s professional duty to ensure compliance, confidentiality and competence remains paramount, there have been certain developments in this field which assign liability to the other members AI supply chain. The European Union, in its New Product Liability Directive (PLD) covers software and AI systems, whether they are embedded in the hardware or distributed independently. It therefore assigns liability to the manufacturers, third-party software developers and intermediaries when their defective AI software causes harm.40 While ultimate liability may still rest with the lawyer
38 John Villasenor, How AI will revolutionize the practice of law | Brookings, Brookings (Mar. 20,
2023), https://www.brookings.edu/articles/how-ai-will-revolutionize-the-practice-of-law/. 39 Riley Kelly, The Power of Training: The Importance of Educating Lawyers on AI Tools to Successfully Scale Pro Bono Efforts - Pro Bono Institute, Pro Bono Institute (Aug. 8, 2024), https://www.probonoinst.org/2024/08/28/the-power-of-training-the-importance-of-educating-lawyers-onai-tools-to-successfully-scale-pro-bono-efforts/. 40 Megan Howarth, AI liability – who is accountable when artificial intelligence malfunctions?, Taylor Wessing (Jan. 7, 2025), https://www.taylorwessing.com/en/insights-and-events/insights/2025/01/ai-liabilitywho-is-accountable-when-artificial-intelligencemalfunctions#:~:text=The%20broader%20remit%20is%20significant,data%20for%20the%20first%20time).
depending on the circumstances, this development shows that as AI becomes more regulated and integrated into our systems, there is a growing possibility of liability being directly attributed to it. Even today, when approved legal AI errs, an argument to assign blame to the AI developer can arise. Maybe there was a bug in the system, maybe the developer did not train the AI sufficiently, or perhaps necessary safeguards were not added. In such cases, the injured party will have to show the precise technical mechanism that failed, and that the developer contractually had a duty of care and that such care has led to reputational, financial or procedural harm. It will have to be demonstrated by the claimant that the risk was foreseeable and preventable, and that no tampering on the claimant's end has caused the error.
However, even with these safeguards the final liability, in this authors opinion , will and should lie in the hands of the legal professional. Machines cannot stand trial or pay damages. They have no moral agency, and it would be counterproductive for victims to have to pursue compensation from a non-human entity in situations such as wrongful arrests or serious accidents. At some stage, humans are responsible for how and why the AI system did what it did. Assigning personhood to AI would risk letting companies off the hook, in effect, encouraging them to disclaim control over the “independent” machine. 41
CONCLUSION
In this author's opinion, the rising tide of AI brings with it a multitude of changes backed by new possibilities and progress. Instead of actively resisting such a change, we as part of the legal professional society must be the pioneers of such change. Yes, ethical problems arise, and certain problems do not have answers, yet these challenges should be viewed as opportunities to strengthen the profession rather than as threats. Moreover, the solutions to today's problems are not far. With proper collaboration from lawyers, policy makers, regulators and developers, we can live in a world where the AI works for us in its best capacity. We must ensure robust frameworks on transparency, accountability and AI oversight to ensure that innovation doesn’t lead to injustice. With this in motion, the legal sector has the ability to not just adapt to the
41 Erica Stanford, Who is responsible when AI acts autonomously & things go wrong?, GLI (May 15,
2025), https://www.globallegalinsights.com/practice-areas/ai-machine-learning-and-big-data-laws-andregulations/autonomous-ai-who-is-responsible-when-ai-acts-autonomously-and-things-go-wrong/.
changing landscape but to lead it. We can shape AI in the way it's meant to be, by ensuring the protection of the moral and ethical foundations of law.