By Sulaksh Editorial Team · Written from the official DU Faculty of Law Sem VIII syllabus (EC 1282) · Reviewed Sept 2026
Dear counsellor of 2031 — that is you, five years from now, reading this guide as nostalgia. Your client is a hospital whose diagnostic AI missed a tumour. The model's trainer is in another country, the deployer followed the manual, the doctor trusted the screen. Everyone behaved reasonably; someone is dead. Who pays? Honours cluster HLP-6 — Artificial Intelligence and Law, Medical Technologies and Law, Corporate Insolvency — is built from questions like that one: disputes the statute book did not foresee, graded on how honestly you reason with thin law. This guide is a set of memos from the future those papers are preparing you for. Argue carefully, admit uncertainty, propose the fix. That is the entire examination technique, stated upfront.
Memo one: the machine with no one inside
The cluster opens with its hardest memo — can an AI system be a legal person? Rights presuppose personhood, personhood presupposes responsibility, and current systems cannot answer for themselves, which is why liability keeps landing on humans: developers, deployers, users. Write the personhood debate as a brief with two funded sides (accountability gaps demand entity-like solutions; anthropomorphism risks smuggling rights where only duties belong) and land on the intermediate instruments — limited agency models, compensation funds, insurance pools. Then the concrete sibling: the DPDP Act, 2023 applied to training pipelines that inhale personal data at scale. Consent, notice, breach duties, children's data — run every AI-data problem through both memos at once (authorship/personhood and data protection) and say you are doing so. Examiners in this cluster reward students who show a dispute's double citizenship. Keep the DPDP rights-and-duties table as your only settled-law anchor; everything else here moves.
Memo two: harms looking for defendants
Civil liability is a seating chart: place every defendant and the music starts. Contract seats the supply chain (who warranted what, from data vendor to deployer). Tort seats negligence for opaque systems, product-liability analogies, and the learned-intermediary question when AI gives professional advice. Consumer law seats the user facing deficient AI services and dark patterns. IP takes the next memo on its own: AI-generated works against the human-authorship requirement, training-data reproduction against fair-dealing's narrow Indian gate, the Thaler/DABUS inventorship litigation the syllabus itself cites — track how jurisdictions diverge and argue India's lane. Criminal justice follows: predictive-policing bias, deepfake offences under the BNS/IT framework, and digital evidence — admissibility, authentication, detection — where procedure outranks theory. Workplace AI closes the set: hiring algorithms, surveillance, gig management, and the governance choice between risk-tiered regulation and innovation-first sandboxes. The method never changes: apply existing doctrine exactly, mark where it strains, propose the repair.
Memo three: the body as a regulated site
Medical technology is HLP-6's oldest memo and its most litigation-dense. Devices, biotechnology, pharmaceuticals: learn each regime as risk, gatekeeper, standard, gap — pre-market approval versus post-market surveillance, and India's honest position between them. Negligence is pleaded, not narrated: duty, breach, causation, every time, with Bolam/Bolitho transplants, Jacob Mathew's criminal-negligence guardrails, and consumer-forum jurisdiction over medical services in the citation kit. New tools stress old standards — AI diagnostics, surgical robots — and the answer is to run the triad on the new facts, not to declare the law obsolete. Then the ethically loaded trio: assisted reproduction under the altruistic-only statutes, organ transplantation with consent and brain-death criteria, research ethics — each answered with law plus one named ethical framework (autonomy, justice, beneficence), because unnamed ethics is just opinion. Telemedicine guidelines and health-data flows under DPDP complete the set. And note the cluster's favourite trick: the AI-diagnosis error is simultaneously a negligence, data-protection, and device-regulation problem. Say all three. Collect from all three mark pools.
Memo four: the machine that actually works — IBC
After three memos of uncertainty, insolvency is relief: the IBC, 2016 examined as a complete, working machine. Learn the origin story seriously (SICA's failure, the committee reports, the clean-slate rationale) because "why IBC exists" justifies everything strict about it — timelines, creditor primacy, limited interference. CIRP is a timeline with teeth: default threshold, Sections 7/9/10 triggers, moratorium scope, CoC voting arithmetic, plan approval, the 330-day outer wall. Examiners set sequence questions — what happens on day X, who votes, what needs 66% — so draw the timeline until it is reflex. Liquidation turns on the Section 53 waterfall, memorised verbatim, plus avoidance transactions. Appeals run NCLAT to Supreme Court with restrained interference; offences punish fraudulent trading. Ratios in one line each: Essar Steel (CoC primacy, timelines), Swiss Ribbons (validity), Vidharbha (admission discretion). IBC answers are scored like arithmetic — thresholds, majorities, days, waterfall order, exact. Get them exact.
Two drills: allocating harm, counting days
Drill one, allocation. Facts: a hiring algorithm trained on ten years of company data systematically downgrades women applicants; the vendor's contract disclaims bias liability; HR followed the scores blindly. Seat the defendants. Vendor: product-liability and negligence arguments, contract disclaimers tested against unfair-terms and consumer law, plus data-protection duties if training data included personal information. Employer-deployer: direct discrimination liability regardless of the tool — delegation to software is no defence — plus DPDP duties for applicant data. HR individuals: contributory process failure. Remedies: compensation, algorithmic audit directions, hiring-process injunctions. State the residual risk expressly (historical bias no party created) and propose the fund-or-insurance fix. Drill two, CIRP date-math. Default on 1 March; Section 7 filed 1 April; admitted 15 April — moratorium runs, the 330-day clock starts ticking; CoC constituted, resolution plans invited; plan approved with 70% vote on day 250; NCLT sanction day 290; dissenting financial creditor challenges. Test each date: admission within the filing discipline, vote above 66%, sanction within the outer wall, challenge grounds limited to statutory non-compliance. Two drills, two reflexes — harm-seating and date-counting — and HLP-6's two hardest question types become procedures instead of panics.
The uncertainty ledger
Keep a running ledger through the term with three columns: settled, straining, silent. Settled: CIRP timelines, waterfall order, DPDP duties, negligence triad — examined for precision, so drill them as closed material. Straining: training-data fair dealing, diagnostic-AI liability, deepfake evidence standards — examined for argued positions, so prepare two-sided briefs. Silent: AI personhood, cross-border health-data enforcement, autonomous-device regulation — examined for design proposals, so prepare institution-and-metric answers. Before the exam, count the ledger: if everything sits in "settled", you have prepared the wrong paper; HLP-6's centre of gravity is the middle column. And in the hall, label your reasoning accordingly — "on settled law…", "the strain point is…", "de lege ferenda…". Latin used sparingly and correctly signals the fluent; the ledger ensures there is substance behind the signal. Counsellors of 2031 are not made by memorising 2026 — they are made by practising, now, the exact reasoning the future will demand daily.
The doctor, the screen, and the standard of care
The single most examinable scene in medical technology is quiet: a doctor follows an AI diagnostic suggestion that turns out wrong. Is the standard of care the reasonable doctor unassisted, the reasonable doctor with such tools, or the tool's own validated accuracy? Run the negligence triad on all three versions and watch the answer change — duty now includes deciding whether to consult the system at all, breach turns on overriding versus deferring, causation tangles around the human-in-the-loop. Add the learned-intermediary question (does the AI vendor owe duties directly to patients, or only through the physician?) and the device-regulation layer (was the tool even approved for this use?), and one clinical vignette touches three units. Prepare this scene as a reusable template with blanks for the technology, because the examiner will keep changing the machine — imaging AI today, surgical robots tomorrow, triage algorithms next. The template outlives every gadget; that is precisely why it is worth building.
Why is insolvency grouped with technology papers? Because the cluster's real theme is law under velocity — regimes that must decide fast under uncertainty. IBC decides company fates in 330 days; AI governance must rule while the technology is still moving. The shared skill is reasoned decision-making with incomplete law, which is also the shared marking rubric. Prepare them as siblings and both improve.
Filing instructions
What if the law is genuinely silent? Say so, then reason by closest analogy and propose the fix. "The statute is silent; the nearest doctrine is X; the better view is Y because Z" is honours advocacy, and concealed uncertainty scores worse than admitted uncertainty with reasoning attached.
IBC or tech — where should my hours go? IBC repays hours at a fixed, generous rate: it is finite, numerical, and heavily repeated. Tech topics repay reading widely but thinly. Fund IBC first; it is the cluster's bank.
Where is the official syllabus? DU's course-structure PDF is under DU & College → Others → LLB (Hons) → Semester 1 → Syllabus on Sulaksh. Verify every topic against its HLP-6 units.