India: the Digital Personal Data Protection Act and AI in law firms
By the ChamberOne team 5 minute read
An associate in an Indian firm pastes a client's employment file into an AI tool to produce a chronology. The file contains names, addresses, salary details and medical leave records. Under India's data protection law, who is responsible for what happens to that data next? The question is not exotic any more. It is a routine part of deciding whether and how to use AI in a law firm.
Key points
- The Act's roles of data fiduciary and data processor shape who answers for data sent to an AI tool.
- Map what personal data goes into each tool before you adopt it.
- Seek written vendor terms on instructions, security, location, breach notice, deletion and audit.
- Professional confidentiality and client policies apply in addition to data protection law.
- Keep a short record of why each tool was approved, and check the official position as provisions commence.
The Act in a sentence, and why firms should care
India's Digital Personal Data Protection Act, 2023 sets out how digital personal data is to be handled, the rights of the individuals it concerns, and the obligations of the organisations that decide how and why it is processed. The Government of India has published material explaining the Act and its rules on its official sites, and the rules have been notified. Parts of the regime commence in phases, so check the current position on the official pages rather than relying on a summary.
Law firms hold personal data of clients, opponents, witnesses, employees and applicants. If an AI tool receives any of it, the question of who is responsible for it is a live one, whatever the size of the firm.
Fiduciary and processor: who is who
The Act uses the term data fiduciary for the party that decides the purpose and means of processing, and data processor for a party that processes personal data on a fiduciary's behalf. A firm that decides to use an AI tool on client files will usually be acting as a fiduciary, and the AI vendor will often be acting as a processor. The Act's text is the authority, and the exact analysis depends on the facts.
That framing has a practical consequence. The firm cannot hand off responsibility by pointing to the vendor. If a processor mishandles data used on the firm's instructions, the firm still has to answer for it, so the contract and the vendor's security matter a great deal.
In some matters the firm's client is the fiduciary and the firm is its processor or service provider. In-house teams and outside counsel should be clear about which role they hold in each matter, because the answer changes what each side must put in writing.
- Decides the purpose and means of processing
- Usually the firm using an AI tool on client files
- Cannot hand off responsibility by pointing to the vendor
- Processes personal data on a fiduciary's behalf
- Often the AI vendor
- Acts under a contract, so its terms and security matter
Mapping what actually goes into the tool
Start with a simple data map for each AI use. Many problems are found at this stage and cost nothing to fix.
- What categories of personal data go in: contact details, financial data, health information, data about children?
- Whose data is it: clients, staff, counterparties, witnesses?
- Why is it going in, and could the task be done with names and identifiers removed?
- Where is it sent, stored and for how long, and is it used by the vendor for anything else?
- Who inside the firm can use the tool, and is each use recorded?
Contract terms to seek from an AI vendor
Because the Act contemplates processing on a fiduciary's behalf under a contract, the vendor agreement is where much of the protection lives. Ask your counsel to review the terms, and look for provisions on the following points. Where the vendor refuses to answer in writing, treat that as information.
- Processing only on your documented instructions, and no use of your data to train the vendor's own models unless you agree.
- Security measures described specifically, not as a general promise.
- Where data is stored and processed, any onward transfer, and the sub-processors involved.
- Notification to you of a personal data breach, with timing and content stated.
- Deletion or return of data at the end of the contract, and how long it remains in backups.
- Help with requests from individuals who want to exercise their rights, and with any regulator enquiry.
- Audit information you can obtain, such as logs of who used the tool and what was sent.
Client confidentiality is a separate layer
Data protection is not the only obligation. Professional confidentiality and privilege apply on top of it, and a tool that is fine for data protection purposes might still be unsuitable for a sensitive matter. Check the guidance from the Bar Council of India, your state bar council and any other regulator on the use of technology and client information.
Tell clients when you use an AI tool on their material, particularly where the engagement terms or the client's own policies restrict what third parties may receive. Many corporate clients now include conditions about AI in their outside counsel guidelines.
In-house teams and outsourcing
In-house legal teams often sit inside a larger organisation that has its own approved tools and data governance. Using an unapproved consumer AI tool for a contract review may breach internal policy even before it raises any legal question. Raise AI use with the data protection lead early, and keep a record of each approved use.
If you outsource document review, contract abstraction or transcription to a provider that uses AI, ask the same questions of the provider, and ask whether it relies on further sub-processors that you have not been told about.
Questions to put to your own team first
Before approaching vendors, ask what people are already doing. Associates under pressure sometimes use free consumer tools on their own initiative, and they will not volunteer it if they expect a reprimand. An amnesty and a conversation uncover real usage. You can then decide which uses to approve, which to replace with a safer tool and which to stop.
Ask also whether the firm's engagement letters and privacy notices describe the use of technology providers accurately. If the notice was written before the firm adopted AI tools, it may now be out of date, and that is easy to correct.
A short practical policy for a small firm
A firm of five or ten lawyers does not need a hundred-page policy. One page can cover the essentials: which tools are approved and for what; what may never be entered, such as sensitive categories of data or documents from restricted matters; a rule to remove identifiers where the task allows; who reviews every output; and who to tell when something goes wrong.
Review the policy when the Act's provisions come into effect in stages or when the vendor changes its terms. Then keep a brief record of the decision for each tool: what you checked, what the vendor said, and who approved it. If a regulator or client ever asks, that record is your best answer.
Sources
This article is general information about the law, legal practice and technology, current when it was published. It is not legal advice. Laws and guidance change: check the sources linked above and take advice on your own circumstances from a lawyer qualified where you are.
Read next
India's new criminal codes: BNS, BNSS and BSA in practice
What India's three new criminal codes mean in practice: the transition rule, key substantive changes, procedural shifts and electronic evidence reforms.
6 minute readGenerative AI guidance for lawyers: how six markets compare
How to find what regulators, courts and bar bodies have said in six markets, and the common themes across them.
5 minute readIndia: what courts have said about AI in legal work
Public policy and judicial caution on AI in India, and what advocates and firm managers should do.
5 minute read