For a few months now, a strange kind of story has been showing up in Indian courtrooms. A lawyer files a petition. The citations look perfect, properly formatted, the right reporter, the right year. The judge goes looking for the judgment to read the reasoning, and it simply is not there. The case never existed. It was invented by a chatbot, and nobody checked.
When I first read about these cases, I assumed they were rare slip-ups. They are not. Over the last year courts have caught fabricated citations again and again, to the point where the Supreme Court in February called the practice misconduct rather than an honest mistake. And in June, the Court did something it rarely does so quickly. It put out a draft set of rules for how artificial intelligence may and may not be used across the judicial system.
That is what this issue is built around. Not the technology itself, but the line the system is now trying to draw around it.
Because the question underneath all of this is older than any chatbot. Who is responsible? A tool can draft, summarise, translate, and search. It cannot be held accountable. A person can. The draft rules say this plainly: AI may assist, but it can never decide, and the human who signs the filing owns every word in it, including the words a machine wrote.
There is something worth noticing here. For most of the past two years the worry was that people were too slow to adopt new tools. Now the worry has flipped. People adopted them too fast, trusted them too completely, and stopped reading. The law is not trying to ban the tool. It is trying to restore the habit of checking.
We have spent recent issues writing about how digital life keeps becoming legal evidence. This is the next turn of the same road. Now the evidence, the drafting, even the translation can be machine-made. And the people who use these systems are discovering that convenience and responsibility do not move at the same speed.