FBR Expands AI Tax Checks to Catch Undeclared Income and Assets

FBR AI tax checks are set to expand across Pakistan, as the Federal Board of Revenue prepares to take its use of artificial intelligence to the next level — widening AI-powered scrutiny of tax returns to catch undeclared income and hidden assets. The move, reported by Arab News on 1 October 2026, is part of a broader IMF-backed reform push aimed at widening the country’s narrow tax base and lifting revenue collection without raising headline tax rates.

The tax authority has already run a pilot of the technology on a limited number of income tax and sales tax returns. Encouraged by the results, it now plans to apply the models more broadly: the system will be used on income tax filings for the tax year that ended in June 2026 and on subsequent sales tax returns, bringing a far larger share of taxpayers under algorithmic review.

From pilot to full-scale rollout

The pilot phase was deliberately kept small, allowing the FBR to test the accuracy of its models and fine-tune the way flagged cases are handled. Officials now believe the technology is ready for a much wider deployment. By extending the checks to the full set of income tax filings for the year ended June 2026, the authority hopes to identify mismatches that would previously have slipped through manual review.

The expansion also signals a shift in the FBR’s enforcement strategy. Rather than relying on random audits or tip-offs, the tax body is moving toward data-driven compliance, in which every return can be screened automatically against independent sources of information.

How the AI cross-checks work

At the heart of the system is a cross-matching engine. The models compare what taxpayers declare in their returns with records drawn from property registries, banking data, vehicle ownership records and other official databases. When the lifestyle or asset footprint suggested by those records does not line up with the income on paper, the return is flagged for a closer look.

Taxpayers whose returns are flagged may face a formal audit. In less serious cases, the FBR may send a compliance nudge — a notice inviting the filer to review and correct their declaration before enforcement action begins.

Key facts

• The FBR is expanding AI-based checks on tax returns to detect undeclared income and assets, as part of an IMF-backed reform push.
• The technology was first tested in a pilot covering a limited number of income tax and sales tax returns.
• The models will now be applied to income tax filings for the year ended June 2026 and subsequent sales tax returns.
• The system cross-checks declarations against property, banking, vehicle and other official records.
• Flagged returns may face audit or receive compliance nudges to correct their filings.
• The campaign carries the slogan \”FBR Sab Janta Hai\” — \”FBR knows everything\”.

A message to filers

The public-facing campaign behind the rollout leaves little room for ambiguity. Under the slogan \”FBR Sab Janta Hai\”, the authority is telling taxpayers that gaps between their declared income and their documented assets are now visible to its systems.

For honest filers, the change should mean little more than the status quo: accurate returns that match third-party records will pass through the screening without incident. For those with undeclared income, the era of hoping discrepancies go unnoticed is drawing to a close.

Tax experts say the success of the programme will depend on execution. The models must be accurate enough to avoid harassing compliant taxpayers with false flags, and the FBR will need to give filers a fair opportunity to explain or correct genuine mismatches. If those safeguards hold, the AI expansion could become one of the most consequential tools in Pakistan’s effort to document its economy — and to make sure that what people earn, own and declare finally add up.