Pakistan restricts govt employees from uploading sensitive data to public AI tools

Government employees have been barred from uploading classified documents, official emails, software source code,  and citizens’ personal information to publicly available artificial intelligence tools.

The restrictions are part of the National Cyber Security Handbook 2026-27, issued by the National Cyber Emergency Response Team under the Pakistan Information Security Framework 2026.

The handbook warns that sharing sensitive information with public AI platforms could expose government and personal data to leaks and unauthorized use.

Under the new guidelines, government officials will only be allowed to use AI tools and platforms approved by their respective departments. Employees have also been instructed to remove names and other sensitive information from prompts and files before submitting them to AI systems.

Officials have further been prohibited from sharing passwords, administrative login credentials, and API keys with AI tools.

The guidelines also prohibit the installation of unauthorized AI extensions and plug-ins on government devices.

Any accidental disclosure of confidential information through an AI tool must be reported immediately to the relevant IT or cybersecurity team.

The handbook also requires officials to review AI-generated content for accuracy and security before using it for official purposes. AI-generated material cannot be used in government work without human oversight.

The new guidelines place the use of AI in government affairs under three key principles: security, privacy, and human oversight.

Also read:

Punjab’s AI Oversight Committee Is a Good Start, Here’s What It’s Still Missing