You type a question into ChatGPT and it answers in seconds. It can draft a reply to a notice, explain a section of the law, or write a formal letter. Then someone tells you this is illegal in Pakistan. Is it?
No. Using ChatGPT in Pakistan is legal. No law bans it, and no court has called it a crime. The state is building policies to promote AI. The problem is never the tool. The problem is what you feed it and what you do with its answers.
What the state says about AI
Pakistan has an official AI policy. The National AI Policy was approved by the federal cabinet in July 2025. It calls for training one million people in AI by 2030 and building a secure AI ecosystem with human oversight. The policy is published by the Ministry of IT and Telecom.
The legal backing came in January 2025. The Digital Nation Pakistan Act 2025 created the National Digital Commission and the Pakistan Digital Authority. The Authority handles digital transformation and data governance, including AI. The full Act is on the Pakistan Digital Authority website.
In February 2026 the Authority announced the Islamabad AI Declaration. It calls for AI that is sovereign, responsible, and accountable, with a human answerable for what the machine does. The declaration is on the PDA website.
So the official direction is clear. The government wants people and businesses to use AI. It is not looking for a reason to ban ChatGPT.
When ChatGPT use becomes a legal problem
The law that covers online conduct in Pakistan is PECA, the Prevention of Electronic Crimes Act 2016. It was amended in January 2025. It does not ban chatbots, but it does punish what people do with them. The amendment is on the National Assembly website. We also explained PECA 2016 in plain language.
- False information. PECA section 26A makes it a crime to spread information you know is false, if it can cause fear or panic. Punishment: up to three years in jail and a fine of up to two million rupees. A wrong ChatGPT answer is not a crime by itself. Posting it as fact can be.
- Defamation. If you publish what the chatbot says about a real person, and it is false and harms their reputation, that is defamation under the Pakistan Penal Code, punishable up to two years. The chatbot did not write it for the public. You published it.
- Deepfakes and intimate images. PECA section 21 covers fake images and videos. Making a deepfake of someone, or sharing intimate images without consent, can mean up to five years in jail. It is worse when the victim is a minor.
- Electronic fraud. Using AI to fake an invoice or a payment instruction for money falls under PECA section 14. Up to two years or a fine of up to ten million rupees, or both.
In every case the person who acted is responsible, not the tool. The investigating agency is the National Cyber Crime Investigation Agency, which replaced the FIA cybercrime wing in 2025.
What you paste into ChatGPT can stay there
Pakistan still does not have a data protection law. The Personal Data Protection Bill has been a draft for years. It is still not an Act.
That matters because a chatbot is a service run by a company outside Pakistan. What you type into it can be stored, read by humans, or used to improve the model.
What protects you is piecemeal. Article 14 of the Constitution says the dignity of man and the privacy of home are inviolable. And PECA section 41 makes it a crime, up to three years, for a service provider to disclose a person's personal information without consent.
The practical rule is simple. Do not paste your CNIC number, your passport, your passwords, or your case documents into a free chatbot. For a sensitive matter, remove names and numbers first.
Even the government restricts how officials use AI
In September 2026 the National CERT issued the National Cyber Security Handbook 2026-27. It bars government employees from uploading classified documents, official emails, source code, and citizens' personal data into public AI tools. Digital Pakistan covered the handbook.
Officials may use only AI tools approved by their department. They must strip names from prompts. Anything AI produces for official work must be reviewed by a human before use.
That is the same rule for you, minus the official part. Keep sensitive data out. Check the output before you act.
Can the government block ChatGPT one day?
It could. Under PECA section 37 the PTA can block online content it considers unlawful. It has blocked gambling and betting apps by the hundreds.
But a block is about access, not about using the tool. Today there is no ban and no move toward one. The risk that matters is not a block. It is what you do with the answer.
Using ChatGPT for legal questions
People use ChatGPT to understand the law, and that is fine. The Supreme Court said so in March 2025. In Ishfaq Ahmed v. Mushtaq Ahmed, reported as PLD 2025 SC 582, the Court ruled that AI can help with legal research, drafting, translation, and case management. It cannot decide a case, weigh evidence, or write a judgment's reasoning. The News covered the ruling.
The Court warned about automation bias and hallucinations, where the AI produces made up information. That warning is not theoretical. AI that invents case law has cost lawyers fines abroad, and the same failure can happen here.
In 2026 the judiciary issued the National Guidelines for the Use of AI in Judicial Institutions. The core rule: a human stays in the loop. The same rule works for you. Use AI to understand your problem and learn the right questions. Do not let it make the decision. Before you trust any legal AI, see what you should verify first.
A short checklist before you paste anything
- Treat the answer as a draft, not as a decision.
- Never paste CNIC numbers, passwords, bank details, or case documents.
- Open the law it names. If you cannot find the section or the case, do not rely on it.
- If you publish its output, check it for defamation and false claims first.
- For money, freedom, or family matters, see a lawyer before you act.
What this means for you
Using ChatGPT in Pakistan is legal. What the law does regulate is content, data, and responsibility.
Keep your private data out of the prompt. Keep a human in the loop. Verify before you act. Do that, and the tool stays a tool.
Browse more legal guides on the Lexiber blog, or start from the home page.
Ask Lexiber free in English or Urdu. It shows you the exact section behind every answer.
