Islamabad: Pakistan is developing an artificial intelligence language model called Buraq, with the government seeking to make digital services more accessible in Urdu and regional languages while building computing capacity within the country.
IT and Telecommunication Minister Shaza Fatima Khawaja outlined the initiative at a World Bank organised panel during the Global AI and Digital Summit in Seoul. The ministry’s October 21 account places Buraq within a broader programme combining national AI infrastructure and local language capability.
What exactly is Buraq?
Buraq is described as a large language model, or LLM: software trained on substantial amounts of language data to recognise patterns and generate responses. Models of this kind can support applications that answer questions, summarise documents, translate text or help draft material.
A model is not necessarily a finished chatbot. It can be the underlying technology used by websites, applications or institutional systems.
For Buraq, those applications remain possibilities rather than a verified feature list. Reporting on the announcement says a public release date and detailed technical specifications have not been announced.
Why Urdu and regional languages matter
The practical opportunity is to reduce the language barrier between citizens and digital services. A person seeking information about an application, a government scheme or an official document should not need advanced English to understand the answer.
Useful local language AI, however, requires more than translating English sentences. Evaluation would need to examine grammar, context, local terminology, regional variation and the Urdu and English mixture common in everyday communication.
The government has not published a detailed list of supported regional languages or independently testable results establishing Buraq’s performance. Nor does the announcement explain whether it is being trained from the beginning or adapted from an existing model.
How identity, payments and government records fit in
The minister linked AI enabled public services to digital identity, payments and secure data exchange. In that framework, NADRA provides identity infrastructure, Raast supports payments, and WASL is intended to enable authorised information exchange between institutions.
These systems perform different jobs. A language model might explain how to apply for assistance; checking identity, establishing eligibility and transferring money require separate systems and lawful authority.
Their mention together does not prove Buraq is already connected to citizens’ records or able to authorise payments.
The World Bank approved USD 78 million for Pakistan’s Digital Economy Enhancement Project in March 2024. That financing supports wider digital public services and should not be presented as Buraq’s individual development budget.
Does keeping data in Pakistan make it safe?
Domestic computing can give institutions greater control over where information is processed. It does not automatically establish privacy, cybersecurity or independence from foreign hardware and software.
Those protections depend on access controls, security testing, retention rules and accountability. Running a model on a citizen’s query is also different from using that query to train it; Buraq’s eventual policy will need to explain both.
The IT ministry’s policy register still labels the National Data Governance Policy 2026 as a draft. An August Pakistan Digital Authority update said consultation feedback would be incorporated before finalisation. These records do not establish that every proposed safeguard is already enforceable.
What will determine its usefulness?
Like other language models, Buraq will need testing for confidently incorrect answers, commonly called hallucinations. Fluency is not proof of factual accuracy.
A credible rollout would make its capabilities and limits visible: who developed it, what data it uses, how performance differs across languages, what it costs and how users can challenge errors.
Buraq’s strongest potential is practical: helping people obtain reliable information in languages they use. Whether it delivers will depend on demonstrated performance and safeguards, rather than its national branding alone.
