AI data-crunching supercharges Inland Revenue tax net and raises the bar for legal advice

Instances that commonly went under-the-radar now easy pickings for Inland Revenue

AI data-crunching supercharges Inland Revenue tax net and raises the bar for legal advice

This article was produced in partnership with Baker Tilly Staples Rodway

The game has changed for tax collection in New Zealand. What once slipped through the cracks of Inland Revenue's oversight now lands squarely in the crosshairs of sophisticated artificial intelligence systems that can analyse millions of transactions in minutes rather than months.

"There's just a lot more activity in the market now," said Andrew Dickeson, director for taxation services at accounting and business advisory firm Baker Tilly Staples Rodway Auckland.

The numbers tell the story of this technological leap. In the first half of the 2024 financial year, IRD opened 3,600 audits – 50% more than the same period the previous year. Through these audits, Inland Revenue found $600m of additional tax that should have been declared, with half of that amount coming from fewer than 10 audits of the country's largest businesses.

For businesses and individuals alike, the message is clear: the days of flying under the radar are over. The combination of AI-powered analysis, international data sharing, and real-time monitoring is creating an environment where Inland Revenue will catch up with more and more wayward practices – ensuring close to perfect compliance is now a realistic goal. The technology allowed Inland Revenue to screen more than three million returns in the most recent financial half, leading to reviews of 30,000 of those returns, with audits, screening and voluntary disclosures adding $859m to tax revenue.

For lawyers, this new reality demands enhanced vigilance, more thorough due diligence, and a rethinking of professional risk management in tax-related matters.

Property sector under the microscope

The property sector has become a particular focus, with IRD uncovering more than $150m in undeclared income tax and GST in this area. The $153.5m discrepancy for the first nine months of the current financial year almost matches the $156.8m figure for the entire 2023-2024 financial year.

"There's two key areas that they appear to be focusing on– they are looking at GST, and property transactions," Dickeson said. Inland Revenue has found $72.9m in discrepancies from property developers alone – a 48% increase on the same time last year.

Advances in technology enable Inland Revenue to catch issues that commonly went under the radar previously. One area involves commercial land sales and whether they can be zero-rated for GST. A buyer can pay full price but claim a component back from IRD as a second-hand goods claim, requiring the lawyer to ensure the seller warrants they are not GST registered. The problem arises when some people are not registered but should have been.

“These cases have gone under the radar previously, and anecdotally, tax practitioners are seeing a jump in these kinds of cases now being caught by IRD,” said Dickeson.

Another area catching purchasers off-guard involves building purpose changes. When a purchaser decides to use part of a property purchased for commercial purposes for residential use instead, no GST has been charged in the commercial sale (zero rated), but the buyer must calculate the portion of building used for residential purposes and pay Inland Revenue a proportionate amount.

"That can come as a bit of a shock to the purchaser because that's on them," Dickeson explained. "They think it was zero-rated and normally they're short on money because they've just paid for the actual underlying land, and then it's 15% of what's probably a reasonably large number that you then have to pay to IRD."

The AI advantage transforms data crunching

The transformation stems from artificial intelligence's ability to process vast amounts of information that would previously have overwhelmed human analysts. In turn, this provides Inland Revenue with a new tool to review positions taken by taxpayers. This can assist Inland Revenue with reviewing transactions that have been zero-rated as an export sale through to the correct tax treatment of a land transaction.

Dickeson explained Inland Revenue has long had potential access to customs data and land records, "but they were swamped".

Before, Inland Revenue had to devote substantial manpower to grind through the data to catch such cases, but now AI can do that work. “The interplay with AI, which can look through basically every transaction of every export happening over a certain period in a few minutes, changes the game entirely.”

The system can now also analyse large amounts of data for certain industries or regions to find out what level of GST a business ought to be paying, identify long-standing errors in tax filing practices, and determine whether there might be a lot of cash transactions occurring.

"They're doing that right now – every time you file a GST return, they're running macro analysis over the data to see whether you're sitting at the profit level that they would expect," Dickeson said.

"There is just a lot more activity going on compared to the old days when tax evasion investigations focused primarily on bank accounts."

Global reach expands enforcement power

The technological enhancement combines with expanded international cooperation to create a formidable enforcement network. IRD now pulls extensive data from overseas, covering people with overseas bank accounts and crypto trading using foreign exchanges. This sophisticated capability stems from agreements with overseas tax authorities, often linked to double taxation agreements.

"Information-sharing obligations on each party lie at the back of those particular agreements," said Dickeson.

Inland Revenue can also get overseas jurisdictions to chase taxpayers for tax owed, rather than waiting until the person returns and catching them at the airport – they can proactively pursue people overseas as well.

The crypto sector provides a prime example of this reach. IRD has identified 227,000 unique cryptoasset users in New Zealand undertaking around seven million transactions with a value of $7.8bn. Inland Revenue has issued 160 warning letters advising customers to declare their income from crypto transactions, which it estimates to be $2.7m in unreported income.

The exponential curve in terms of data-crunching ability shows no signs of slowing. What began as traditional tax evasion investigations looking at people's bank accounts has expanded to examining data from all government departments and overseas sources.

"Back in the old days, it was like tax evasion and looking at people's bank accounts, whereas now it's looking at all government departments pulling in data from overseas," Dickeson said. "And guess what? It's just going to continue the exponential curve in terms of data-crunching ability."

Implications for legal practice

For legal practitioners, this technological shift demands a fundamental reassessment of risk management and client advice protocols. The margin for error has essentially disappeared – what might once have been overlooked minor compliance issues now trigger automatic detection and enforcement action.

Lawyers need to be far more thorough in their due diligence, particularly around GST registration status and property transaction structures. The enhanced international data sharing also means lawyers advising on offshore structures or transactions must consider that these arrangements will likely be detected and scrutinised. Professional indemnity insurance considerations may need revisiting as the risk profile of tax-related legal work has fundamentally changed.

The shift represents more than just technological advancement – it's a fundamental change in how tax compliance operates, with direct implications for legal practice standards and professional liability.