Artificial Intelligence Use Policy for Legal Practice & Litigation

AI Use Policy – JP Lotus
JPL
JP Lotus
Advocates & Solicitors  ·  lotuslaws.in
Internal Policy Document

Artificial Intelligence Use Policy
for Legal Practice & Litigation

Effective DateJune 17, 2026
Issued ByJaskaran Singh, Advocate
Version3.0
Review CycleAnnual
Applies ToAll Partners, Associates & Staff
01
Why This Policy Exists

Artificial Intelligence tools are becoming part of everyday legal work — from research to drafting to document review. Used well, they save time and improve quality. Used carelessly, they can destroy client trust, breach confidentiality, and put a lawyer's licence and a client's case at serious risk.

JP Lotus is issuing this Policy for the following reasons:

  • To protect client confidentiality. Anything typed into an AI tool can potentially be stored, viewed, or used by the AI provider unless we take active steps to prevent it.
  • To protect the firm and our lawyers from professional misconduct risk. AI tools can fabricate case law, misstate facts, and produce confident-sounding errors that can mislead a court if not checked.
  • To comply with our duties of confidentiality and privilege. Litigation involves some of the most sensitive information a client will ever share. That duty does not pause because a tool is "just helping with drafting."
  • To create one consistent standard so that every person at the firm — partner, associate, paralegal, or intern — follows the same safe practices, rather than everyone deciding for themselves.
02
What Is AI?

For the purpose of this Policy, "AI tools" means any software using machine learning or language models to generate, summarise, translate, or analyse text, documents, images, or data. This includes:

  • Large language models (e.g., ChatGPT, Claude, Gemini, Copilot)
  • AI-assisted legal research platforms (e.g., Harvey, Lexis+ AI, Westlaw AI, CaseMine, SCC Online AI features)
  • AI document drafting, review, or contract analysis tools
  • AI transcription, translation, or summarisation services

Basic spell-check or autocomplete is not covered unless it has generative AI features built in.

03
Mandatory Rules — No Exceptions

The following rules apply to every single person at JP Lotus, on every single use of AI, with no exceptions.

1
Turn off AI training/data-sharing before you use any tool. Most AI tools have a setting (often on by default) that allows your input to be used to train their models. Before using any AI software for the very first time, you must locate this setting and switch it OFF. If a tool does not allow you to disable this setting, you may not use it for any firm or client work.
2
Never use an AI tool without logging in. Logged-out / anonymous / "guest" sessions on AI tools must never be used. A logged-in account provides control over data settings. Such control over the data settings are not available while using logged-out / anonymous / "guest" sessions.
3
Never input sensitive client data into an AI tool. No client name, case number, contact detail, financial information, document, or fact pattern that could identify a client may be typed or pasted into an AI tool — unless anonymised first or unless written client consent has been obtained.
4
Client consent is required before entering any personal client data. If a matter genuinely requires personal or case-specific client data to be used in an AI tool, you must first obtain the client's clear, written consent.
5
Permanently delete chats once the work is complete. Once you have finished using an AI tool for a task, you must delete the chat/conversation history from your account, do not just "archive" or "hide" the chats. Do not let old client-related chats sit in your account history when no longer required.
6
No data — of any kind, ever — may be used to train an AI model. This is the single most important rule in this Policy. Whatever the tool, whatever the task: client data must never become training data for any AI system, public or private. This is non-negotiable.
If You Are Not Sure How to Check These Settings

Ask for help. Do not guess, and do not proceed "for now" intending to fix the settings later. The settings must be confirmed correct before you type anything.

04
Responsible Use of AI in Litigation

Litigation carries unique risks because the output of our work goes directly before a court, opposing counsel, and the client. Responsible use in litigation means following a simple discipline:

  1. Anonymise first. Before using AI for research or drafting support on a case, replace client names, identifying numbers, and unique facts with placeholders (e.g., "Client A," "the Petitioner," "the disputed property") wherever the task allows it.
  2. Use AI only for support tasks, never for final judgment. AI may help you find a starting point, summarise a long judgment, or generate a first-draft structure. It must never be the source of the final legal position, strategy, or argument that goes to court.
  3. Verify every citation independently. Every case name, citation, statute, and quoted passage suggested by AI must be checked in an authoritative legal database (SCC Online, Manupatra, official court websites) before it appears in any document filed or sent to a client.
  4. Human involvement at every stage. A qualified lawyer must review, edit, and take ownership of anything AI-assisted document or information before it is shared further.
The Golden Rule of AI in Litigation

AI can help you think faster. It can never think for you, and it can never know your client's case better than you do. Treat every AI output as an unverified first draft — promising, but never final.

05
Do's of Using AI in Litigation
✓ Do
  • Disable the AI training/data-sharing setting before first use
  • Log in through your firm-issued, approved account only
  • Anonymise client names and identifying facts wherever possible
  • Use AI to summarise long judgments, orders, or transcripts as a starting point
  • Use AI to find a first-draft structure for routine drafts (e.g., a standard application format)
  • Independently verify every case citation in an authoritative database
  • Get written client consent before entering any personal client data
  • Delete the chat history permanently once your task is complete
  • Report anything that looks like a data leak or unusual AI behaviour immediately
✗ Don't
  • Don't use any AI tool with default training settings left ON
  • Don't use AI tools in guest mode / without logging in
  • Don't paste a client's name, case number, or documents directly into AI
  • Don't rely on AI-suggested case law without checking it yourself
  • Don't submit any AI-drafted content to a court without full lawyer review
  • Don't use AI to decide litigation strategy, settlement value, or final arguments
  • Don't leave old chats sitting in your AI account after work is done
  • Don't use free, unapproved, or personal AI accounts for case work
  • Don't assume consent — always get it in writing before entering personal data
06
Permitted vs. Prohibited Litigation Tasks
TaskStatus
Summarising a judgment or order for internal understandingPermitted
Drafting a first-draft skeleton of a routine applicationPermitted
Legal research on a point of lawCaution
Drafting a final affidavit, plaint, or petition using AI without reviewProhibited
Pasting case files, client documents, or evidence into AIProhibited
Using AI to predict case outcome or settlement strategyProhibited
Using AI to translate or summarise client communicationCaution
Submitting AI-generated content to court without verificationProhibited
07
Client Consent Requirements

Client personal data may only be entered into an AI tool if all of the following conditions are fulfilled:

  • The client has given clear, written consent specifically for the use of AI tools with their data;
  • The consent explains, in plain language, what data will be used and for what purpose;
  • The AI tool being used has its training/data-sharing setting confirmed OFF (per Section 3, Rule 1);
  • The use is logged in through an approved, firm-issued account (per Section 3, Rule 2); and
  • The chat will be permanently deleted once the task is complete (per Section 8).
No Consent, No Input

If a client has not given written consent, their personal or case-identifying data must not be entered into any AI tool — anonymise the data instead, or do not use AI for that task.

Clients who specifically instruct that no AI be used at all on their matter, such instructions must be fully respected.

08
Data Deletion Protocol

Once an AI-assisted task is complete, the following steps are mandatory:

  1. Close out and permanently delete the chat or conversation from the AI tool's interface (not just close the browser tab).
  2. Where the tool offers a "delete history" or "clear data" function beyond simple archiving, use it — archiving is not deletion.
  3. If the tool retains data for a fixed period before permanent deletion (common with many providers), note this and avoid re-entering the same client data in future sessions until that retention period has fully lapsed.
  4. Do not export, download, or forward AI chat logs containing any client information for "reference later." Take your verified output and discard the chat.
09
Approved AI Tools

Staff may only use AI tools that have been reviewed and approved by firm management, with confirmed enterprise/business-tier settings that allow training data to be switched off.

  • Do not use free consumer AI tools (e.g., free-tier ChatGPT, consumer Gemini) for any firm or client work — these typically lack the controls required by this Policy.
  • Any tool used must comply with applicable data protection laws, including the IT Act, 2000 and the Digital Personal Data Protection Act, 2023, where applicable.
10
Professional Responsibility — You Are Always Accountable

AI is a tool, not a lawyer. The professional and ethical responsibility for every piece of work product remains entirely with the lawyer who supervises or signs off on it.

  • Every document, submission, or advice that uses AI assistance must be reviewed, edited, and approved by a qualified lawyer before it is sent to a client or filed with any authority.
  • A lawyer cannot delegate professional judgment to an AI tool. Competence, diligence, and integrity remain personal obligations.
  • Bar council rules, court rules, and applicable laws continue to apply. AI use does not create any exemption or excuse.
  • If you are unsure whether a particular use of AI is appropriate, stop and ask a senior partner before proceeding.
11
Pre-Use Checklist

Run through this checklist every time, before you type anything into an AI tool:

Before You Use Any AI Tool
Is the AI training / data-sharing setting switched OFF? Confirmed, not assumed.
Am I logged in through my firm-issued, approved account?
Have I removed or anonymised all client-identifying information?
If personal client data must be used — do I have written client consent on file?
Will I verify every fact, citation, and figure the AI gives me before relying on it?
Will I permanently delete this chat once my task is complete?

If you cannot tick every box, stop and do not proceed until you can.

12
Environmental Impact

AI use carries a well-documented environmental cost. Please be mindful and make sure you understand that impact before choosing to use any approved AI tool for work.

13
Policy Review

AI technology and the legal landscape around it are both evolving rapidly. JP Lotus will review this Policy at least once a year, and sooner if there are significant developments in law, regulation, or technology that require it.

Questions, concerns, or suggestions regarding this Policy should be directed to Jaskaran Singh.

Issued & Authorised By
Jaskaran Singh
Advocate
JP Lotus  ·  lotuslaws.in
Effective: June 17, 2026  |  Version 3.0  |  Review Due: June 2027  |  This is an internal policy document of JP Lotus.