LW2191 Foundations of Law: Tasks 1 & 2 Combined Assessment

Student Number: [YOUR STUDENT NUMBER]

Module: LW2191 Foundations of Law Advanced

Assessment Title: Reflective Log on AI-Generated Client Letter and Ethical Issues Raised by Generative AI in Legal Writing

Tasks:

Task 1: Reflective critique of AI-generated client letter

Task 2: Essay on ethical issues raised by Generative AI in legal writing

Word Count (Tasks 1 & 2 combined, excluding footnotes and bibliography): 1000

Submission Date: 7 January 2026

Task 1: Reflective Critique of AI-Generated Client Letter

Content Accuracy & Authorities

What the AI letter lacks is critical legal grounding though structurally distinct dispute resolution analysis is offered. The letter introduces mediation, arbitration and litigation as more or less balanced to no guarantee of contemporary English civil justice policy: ADR has become institutionalised as no longer an option. The explicit authority to direct ADR participation is now in place in courts (Churchill v Merthyr Tydfil County Borough Council [2023] EWCA Civ 1416] and marks the transformation of the voluntary model of Halsey v Milton Keynes General NHS Trust [2004] EWCA Civ 576).

The letter does not refer to Sale of Goods Act 1979, s. 13 14(2A), which define the solid legal argument by GCC Limited (batteries not fitting specifications = breach of condition). It is noteworthy that the letter has not comprehended the dispute as one that is legal (under breach of contract, remedies), but one that is purely commercial (preservation of reputation, Chinese contracts). Modern research (Ahmed 2024; Corts 2023) proves that ADR is coursed by pre-action procedures and regulated pathways of referrals; the letter introduces ADR as an option not an agency requirement.

Authorities wholly absent. Required citations: Halsey [2004] EWCA Civ 576; Churchill [2023] EWCA Civ 1416; Sale of Goods Act 1979, ss. 13, 14(2A); Masood Ahmed, 'Revisiting compulsory alternative dispute resolution in the English civil justice system' (2024) 43(4) Civil Justice Quarterly 265; Pablo Corts, 'Embedding alternative dispute resolution in the civil justice system' (2023) 43(2) Legal Studies 312; Stefan Fafinski & Emily Finch, Legal Skills (10th edn, OUP 2025) ch. 11.

Recommendation Assessment

Mediation advice is both commercially reasonable and legally missing. The letter never discusses ramifications of the denial of ADR: in Halsey, the courts can punish with costs or in Churchill, requirements for the use of ADR. The good legal stance of GCC Limited (implied conditions breach) is not addressed; resolving by mediation may lead to compromise that would otherwise not be needed in court.

Even modern case law and civil procedure now anticipate pre-trial ADR. This may lead to ordering of adverse costs where one side prevails despite not attempting ADR. This judicial expectation is absent in the letter so creating a false picture of the actual cost-benefit analysis of the recommended option.

Structure & Format

Professional formatting: Right letterhead, time, address to client, salutation, subject matter, rational clear sections, professional manner, right concluding. Nevertheless, the letter is not legalistic: it regards dispute resolution as commercial in nature without setting up the legal status of GCC Limited first. A more robust framework would: (i) give facts and legal problem; (ii) describe rights of law and strength of claim under Sale of Goods Act 1979; (iii) thereafter analyze how each of the mechanisms conforms to the rights and commercial interests.

Task 2: Ethical Issues Raised by Generative AI in Legal Writing

Core Ethical Concerns

There are five overlapping ethical risks of generative AI in legal writing: (1) hallucination and fabrication; (2) lack of transparency and accountability; (3) bias and fairness; (4) confidentiality risks; (5) unreliability as authority.

Hallucination and cheverria. AI systems produce convincing-sounding yet entirely imaginary case citations, laws and legal propositions. In Mata v. Lawyers at Avianca Inc. used false case references made on ChatGPT and were fined and sanctioned by the court. The court in R (Ayinde) v London Borough of Haringey; Al-Haroun v Qatar National Bank QPSC [2025] EWHC 1383 (Admin) indicated that they saw material with indicators of an AI being used but which had not been, and that professional responsibility is not delegable, since it was an AI driver, not the driver of the car. counsel have personal responsibility to all the powers put before this court.

Accountability and transparency. Courts have become particularly demanding in terms of reporting the use of AI and certifying all data produced by AI. According to the American Bar Association Model Rules, the application of AI by lawyers requires their competence and diligence; responsibility cannot be assigned to a tool.

Bias and Fairness. Hit and Miss AI systems can create problems with systemic inequity. Algorithms used in predictive policing are biased against racial groups; AI-based tools working with sentencing data can make recommendations aimed at a more stringent result of a certain group of defendants. This goes against the obligation of a lawyer to be just.

Confidentiality Risks. To add sensitive client data in third-party AI systems (e.g., ChatGPT) is potentially a violation of professional responsibility and data protection and confidentiality obligations. Attorneys can accidentally reveal sensitive information in cases, client information, or legal plan to AI vendors or other users of data available on the platform.

The Plausibility Compensation. This is the most pernicious possible ethical challenge: the AI produces authoritative and professionally written content, but cannot be pointed to as lacking the evidence base that defines a valid legal advice. The AI-generated letter to the client analysed above illustrates this threat.

How These Issues Can Be Addressed

1. Compulsory Check-up Processes. Lawyers should implement binding rules: any AI-generated reference is to be processed separately by checking such sources against primary ones. This is already a professional requirement according to the ABA rules and UK Bar Standards; the introduction of AI does not reduce it.

2. Transparency and Disclosure. Lawyers should educate customers regarding AI usage, its restrictions, and confidentiality concerns. Jurisdictions ought to explain the compulsory disclosure provisions. This system is reflected in the traffic-light system (Green/Amber/Orange categories) used in the University of Leicester: students have to use AI and be monitored.

3. Impact Assessment and Bias Auditing. Firms need to audit AI tools before rolling out their deployment, they should be audited regarding bias between demographics and the type of case they handle. Bias testing procedures must become the norm, and applications that entail a higher risk (predictive analytics, sentencing prescriptions) must get more scrutiny.

Conclusion

Generative AI is efficient in legal writing and research. Nonetheless, these advantages are compromised by unaddressed ethical risks: hallucinations, partiality, breach of confidentiality, and lack of accountability. Integrity and candour constitute the initial obligation of the legal profession, and they could not be combined with the use of AI without stringent checking, reporting, and supervision.

Bibliography

Amin I and Amin I, Client Beware: The Utilization of Artificial Intelligence Platforms and the Potential Waiver of Attorney-Client Privilege (TALG, November 10, 2025)
Berk RA, Artificial Intelligence, Predictive Policing, and Risk Assessment for Law Enforcement (2020) 4 Annual Review of Criminology 209
Jamesju, ABA Ethics Rules Related to Generative AI (Thomson Reuters Law Blog, November 10, 2025)
Legal Skills - Paperback - Emily Finch, Stefan Fafinski - Oxford University Press
Masood Ahmed, 'Revisiting compulsory alternative dispute resolution in the English civil justice system' (2024) 43(4) Civil Justice Quarterly 265
Pablo Corts, 'Embedding alternative dispute resolution in the civil justice system: a taxonomy for ADR referrals and a digital pathway to increase the uptake of ADR' (2023) 43(2) Legal Studies 312
Pascale Lorber, 'Generative AI, law schools and assessment: where next?' (2025) 59(3) The Law Teacher 267
R (on the application of Ayinde) v London Borough of Haringey; Al-Haroun v Qatar National Bank QPSC [2025] EWHC 1383 (Admin)
Sharp V and others, Judgment of the High Court of Justice, Kings Bench Division, Divisional Court (2025)