Last updated: 12 November 2026
Using ChatGPT or a similar tool to draft an e-mail, summarise a report or write a formula is now routine in Indian offices. The mistakes that cause trouble are rarely dramatic. They are small habits: pasting a client’s contract into a public chatbot, forwarding an answer without checking it, using a generated picture of a real person. This guide goes through eight common mistakes and gives the correct position under Indian law as of the date above.
One framing point first. India does not have a single AI statute. The Ministry of Electronics and Information Technology issued the India AI Governance Guidelines on 5 November 2025, which prefer voluntary measures and the use of existing laws over a new Act. That means the rules that already exist (the IT Act, 2000, the data protection law, copyright law, contracts and professional duties) apply to your use of AI tools today.
Learn this properly. Lexfolio offers AI at Work: Using ChatGPT and AI Tools Safely and Legally, a self-paced certificate course on using ChatGPT and other AI tools safely and legally at work. Course price: Rs 299.
Mistake 1: Pasting confidential or personal data into a tool you have not checked
This is the most common and the most costly mistake. Client contracts, salary sheets, medical details, unpublished results, source code and customer lists often carry a duty of confidence under an employment contract, a non-disclosure agreement or a professional rule. Typing them into an external tool is a disclosure to a third party, whatever the tool’s friendly interface suggests.
The legal pressure points are these:
- Contract. Most employment and client contracts forbid sharing confidential information without permission. Breach exposes you or your employer to a claim.
- Section 72A, IT Act. Disclosing personal information obtained under a lawful contract, with intent to cause wrongful loss or wrongful gain, is punishable with imprisonment up to three years or a fine up to Rs 5 lakh, or both. Intent is the key element, so careless pasting is not automatically a crime, but it is the type of facts from which disputes start.
- Section 43A, IT Act. A body corporate handling sensitive personal data must keep reasonable security practices, or it may be liable to pay compensation to the person affected. This section continues to apply until section 44(2) of the Digital Personal Data Protection Act, 2023 comes into force on 13 May 2027.
- DPDP Act, 2023 and Rules, 2025. From 13 May 2027 the main duties of a data fiduciary are scheduled to apply, including reasonable security safeguards and breach intimation. The Act makes the fiduciary responsible for processing done on its behalf, which will include data sent to a vendor’s tool. Penalties in the Act’s schedule go up to Rs 250 crore for failure to take reasonable security safeguards.
The practical rule: use only tools your employer has approved, check the plan’s terms on whether your inputs are kept or used to improve the service, and mask names, numbers and identifiers when you are unsure.
Mistake 2: Treating the answer as a verified fact
Generative tools produce fluent text, not guaranteed truth. They can invent numbers, rules and even court decisions. The Indian courts have already reacted. On 17 February 2026 a Supreme Court bench led by the Chief Justice of India, Surya Kant, remarked that some pleadings appeared to have been drafted with AI platforms without adequate human scrutiny, and it noted a cited precedent (“Mercy vs Mankind”) that does not exist. In the case of Gummadi Usha Rani v. Sure Mallikarjuna Rao the Court said on 27 February 2026 that a decision relying on non-existent, fake judgments is not a mere error in decision making but could be treated as misconduct. The subject there was courts, but the lesson is general: whoever puts a result into a document or a decision answers for it.
The correct habit is to treat AI output as a first draft by a fast, careless assistant. Check every figure against the source, every legal rule against the bare Act or a trusted database, and every quotation against the original.
Mistake 3: Believing “there is no law on AI, so anything goes”
There is no standalone law, but there are many applicable ones. Fake or misleading content can raise issues of cheating (sections 318 and 319 of the Bharatiya Nyaya Sanhita, 2023), identity theft or personation online (sections 66C and 66D, IT Act), and defamation. Data rules apply to personal data fed to a tool. Consumer protection law applies to misleading claims made to customers, whether a person or a tool wrote them. The 2025 MeitY guidelines also encourage organisations that deploy AI to set up accessible grievance redressal mechanisms.
Mistake 4: Assuming you automatically own, and may freely use, whatever the tool produces
Copyright law deals with this unevenly. Section 2(d)(vi) of the Copyright Act, 1957 says that for a computer-generated literary, dramatic, musical or artistic work the author is the person who causes the work to be created. How far that covers text or images produced by a generative tool from a short prompt is not settled by a court judgment in India. The Department for Promotion of Industry and Internal Trade (DPIIT) released a working paper on generative AI and copyright on 8 December 2025 proposing a “hybrid model” for the use of copyrighted works in training; it is a consultation document, not law.
So: do not assume exclusivity over generated material, do not claim a generated report as wholly your original work where your employer or client expects human authorship, and read the tool’s terms on ownership and permitted commercial use. Never feed in material you are not licensed to copy.
Mistake 5: Making synthetic images, voices or videos of real people
A generated photograph of a real person, a cloned voice or a “face swap” can be misleading or harmful even when made as a joke. Depending on the facts, it can fall under section 66E (violation of privacy, up to three years or fine up to Rs 2 lakh), section 66D (cheating by personation), the obscenity sections 67 and 67A, or defamation. The IT Rules were amended by a notification of 10 February 2026, effective 20 February 2026, to require that permitted synthetic content be prominently labelled and to shorten the time in which platforms must act on government or court orders to three hours. Those duties fall mainly on platforms, but a company that publishes unlabelled synthetic content invites a takedown and reputational damage. Get written consent from any real person whose likeness or voice you use.
Mistake 6: Letting a tool decide about people
Using software to rank candidates, flag “low performers” or set pay without a human reviewing the output carries two risks. First, errors or bias in the data can lead to unfair decisions that an employee can challenge under the employment contract and service law. Second, the personal data used must be handled in line with the data protection law once it applies. The safe course is to let the tool assist, require a named human to decide, keep a record of the reasons, and tell staff and applicants what the tool does.
Mistake 7: Using personal accounts for office work
When employees use private accounts of public tools, the employer has no record of what left the building and cannot prove compliance. The CERT-In directions of 28 April 2022 expect covered entities to retain logs of ICT systems for a rolling 180 days within India and to report listed incidents, including data leaks, within six hours of noticing them. A leak through a personal account is much harder to detect and reconstruct. A company approved route with an access list and logging is better than a ban that staff quietly ignore.
Mistake 8: Not disclosing AI use where disclosure is expected
Clients, universities, regulators and employers increasingly ask whether AI was used. If a contract, assignment rule or submission policy requires a declaration and you stay silent, the problem is no longer the tool but the omission. Where no rule exists, a short line such as “first draft prepared with an AI tool and reviewed by me” is often the honest and safe course. For lawyers and other licensed professionals the duty to verify is stricter: the Supreme Court has said parties cannot rely on AI-generated material and then simply apologise, and the Bar Council of India has told Parliament that advocates cannot escape liability by blaming technology.
A short routine you can use tomorrow
- Classify what you are about to paste: public, internal, confidential or personal data.
- Check that the tool is approved for that class of information.
- Mask names and identifiers where possible.
- Verify each fact, number and citation before use.
- Record the tool and the checks in the file for important work.
- Disclose where a rule or client expects it.
Learn this properly. Lexfolio offers AI at Work: Using ChatGPT and AI Tools Safely and Legally, a self-paced certificate course on using ChatGPT and other AI tools safely and legally at work. Course price: Rs 299.
Frequently asked questions
Can my employer legally ban ChatGPT at work?
A: Yes. An employer can set reasonable instructions on which tools may be used and for what, and breach can lead to disciplinary action under the service rules or contract. A written policy makes the rule clear and fair.
Is it illegal to use ChatGPT to write my assignment, report or article?
A: There is no general statute against it. What matters is the rule of the institution or client, the duty to be honest about authorship, and copyright in any material the tool reproduces. Check the specific policy.
Who is responsible if the AI tool gives wrong advice and my company relies on it?
A: Generally the company that used the output and presented it to others, as the tool provider will usually limit its liability in its terms. Verification and human review are therefore your protection.
Does the DPDP Act already apply to AI tools?
A: The Act and Rules were notified in 2025, but the main obligations of data fiduciaries are scheduled to apply from 13 May 2027. Until then, the IT Act (including section 43A) and contractual duties govern, and it is sensible to prepare now.
This article is general information, not legal, financial or tax advice for any specific case. Lexfolio is a learning resource and does not provide legal services.
Useful books for this topic
- Bare acts of the new criminal laws (BNS, BNSS, BSA)
- Code of Civil Procedure bare act
- Legal drafting and conveyancing books
- Constitution of India (coat pocket edition)
As an Amazon Associate, Lexfolio earns from qualifying purchases. Prices and availability are shown on Amazon.
Learn this properly, with a certificate. Police and FIRs, consumer and bank fraud complaints, property, wills and inheritance, work rights, family law, cyber fraud and RTI, in plain English. Certificate Course in Everyday Law for Every Indian, ₹349: self-paced lessons, module quizzes and a final test; pass at 60% for a verifiable Lexfolio certificate.
Leave a Reply