Many lawyers first encountered AI through ChatGPT. It is a tempting starting point for drafting a petition, summarizing a clause, or strengthening an argument. But is ChatGPT really enough in a field like law, where every sentence must rest on a source? In this article we compare general-purpose AI with law-specific AI through the lens of a lawyer's daily work.
What is ChatGPT useful for in law?
General-purpose models are genuinely strong at working with language. They offer practical benefits for lawyers:
- Simplifying, formatting, or changing the tone of a text.
- Producing a quick first draft on a general topic.
- Summarizing a long text or breaking it down point by point.
- Discussing an idea from different angles and brainstorming.
These tasks are where general-purpose AI saves time. But they all share one feature: the output does not need to be tied to a verifiable legal source. Once a source is required, the picture changes.
ChatGPT's four limits in law
1. Fabrication (hallucination). General models can produce statutory provisions and court decisions that do not actually exist. In law, this means a lost case or disciplinary liability. The article where we cover this topic in depth: Is AI Safe in Law?
2. It does not cite sources. A general model produces its answer from what it "remembers," not from a real database. This is why you have to verify every claim by hand, one by one, which takes back most of the time it saved you.
3. Recency and locality. Turkish law changes frequently. A general model's training data may not include current legislation and may not have a firm grasp of Turkish case law.
4. Privacy. Entering sensitive client information into a general tool whose data handling is not transparent is risky in terms of KVKK and professional secrecy.
What does law-specific AI offer?
Law-specific tools are built to solve these four limits by design. Their most important difference is that they tie every claim they produce to a real statutory provision and court decision, and let you verify that source with a single click. On top of this, they work from current legislation and case law, handle your data in compliance with KVKK, and let you produce and work on documents rather than merely chat.
ChatGPT versus law-specific AI: a comparison
| Criterion | General-purpose AI (ChatGPT) | Law-specific AI |
|---|---|---|
| Sources and citations | Does not cite sources, may fabricate | Every claim tied to a real provision/decision, clickable |
| Recency | Limited by training data | Based on current legislation and case law |
| Privacy | Often unclear | KVKK compliant, not used for training |
| Way of working | Chat box | Live documents, file management, teams |
| Turkish law | General, superficial | Specific to Turkish legislation and case law |
When should you use which?
General-purpose AI is a practical assistant for writing, summarizing, and formatting tasks that do not require a source. But for research, petitions, and contracts that rest on legislation and case law, that is, work that carries legal responsibility, you need a law-specific tool whose sources are verifiable. In practice many lawyers use both together; the critical difference is that legal output always rests on a verifiable foundation.
For an overview of where AI is used in law: The Use of AI in Law.