IslamabadPakistan’s Federal Board of Revenue will deploy AI to cross-check bank, property and customs records against tax filings in fiscal 2027.
Federal Board chairman Rashid Langrial said the system aims to minimize human interaction and detect false returns.
Final proposals go to the budget announcement.
Sources: Dawn
THE NARROW BASE
Pakistan has one of the lowest tax-to-GDP ratios in the developing world — roughly 9-10%, against an IMF target of 15%. Out of 240 million people, fewer than 5 million file income tax returns, and a smaller fraction actually pay. The state runs on borrowing because it cannot tax.
WHO ACTUALLY PAYS
Salaried employees in the formal sector have tax withheld at source — they cannot evade. Agriculture, which is roughly a fifth of GDP, is constitutionally a provincial subject and effectively untaxed at the federal level. Retailers and wholesalers contribute a fraction of their economic weight. The burden falls on those who cannot hide.
WHY AI, WHY NOW
The FBR has tried risk-based audits, third-party data matching, and amnesty schemes for two decades. Each ran into the same wall: tax officers negotiated settlements with filers, and discretion became a market. Removing the human from the loop is the explicit pitch — algorithmic detection is harder to bribe than a clerk.
THE REGRESSIVE RISK
AI cross-matching works best where digital records exist: bank accounts, registered property, customs declarations. The undocumented economy — cash retail, informal real estate, agricultural income — remains invisible to the algorithm. The risk is sharpening enforcement on the already-visible salaried and small-business class while the larger evaders stay outside the data.
THE IMF CONDITIONALITY
Pakistan's $7bn Extended Fund Facility, approved in 2024, made tax-base expansion a structural benchmark. Each tranche release depends on FBR meeting collection targets. The AI rollout is less a domestic reform than a deliverable for Washington — which is also why announcements precede implementation by years.