What Happens When AI Writes a Contract? A Beginner's Guide to Contract Formation

Ask an AI tool to draft a non-disclosure agreement and it will produce something that looks, structurally, exactly like a contract — recitals, defined terms, numbered clauses, a signature block. What it won't tell you is whether the document it produced actually functions as a legally enforceable contract for your situation. That question doesn't go away just because a machine did the typing.

The four requirements haven't changed

Under Indian contract law, an enforceable contract needs an offer, an acceptance of that offer, consideration flowing between the parties, and an intention to create legal relations, between parties who are competent to contract. None of these requirements care who or what drafted the document. A beautifully formatted AI-generated agreement that both parties never actually agreed to is not a contract. A messy one-paragraph email exchange where both parties clearly agreed to specific terms can be.

Where AI drafting actually goes wrong

• Boilerplate that doesn't match jurisdiction — clauses drafted for a different legal system's default rules, dropped into an agreement governed by Indian law.
• Internal inconsistency — a definitions section that contradicts how a term is actually used three clauses later, because the draft wasn't generated with the whole document in view.
• Missing the actual deal — polished language around a structure that doesn't reflect what the parties actually negotiated.
• Overconfident certainty language — clauses stated as though settled law when the underlying question is genuinely unresolved.

What to actually do with an AI-drafted contract

Use it as a first draft, not a final document. Read every clause against what the parties actually agreed, not against what sounds legally correct. And treat any clause that states a legal proposition — what happens on breach, what law governs, what remedies are available — as something to verify against the actual statute or precedent, not something to trust because it reads confidently.