If you think the South African Revenue Service sees only the information you put on your tax return, think again.
One of the more interesting insights from Commissioner Dr Johnstone Makhubu’s launch of the 2026 tax filing season was just how much better SARS has become at connecting information that already exists across the government.
The result is a much more complete picture of taxpayers’ affairs.
Instead of relying mainly on what individuals and businesses declare, SARS is increasingly using artificial intelligence and advanced data analytics to compare information from multiple sources. That allows it to identify taxpayers whose declared affairs simply don’t line up with the economic activity it is seeing elsewhere.
For honest taxpayers, that should mean fewer unnecessary interventions and a smoother experience. For those who deliberately stay off the tax radar, however, the net is becoming considerably tighter.
Following the money
Perhaps the clearest example is the way SARS is now using procurement payment information from National Treasury.
Every week, it receives data showing which businesses have been paid by the government. That information is matched against SARS’s records to identify companies with outstanding tax debt, businesses that should be registered for VAT but are not, and taxpayers whose affairs warrant a closer look.
“We are now able… to look at those people that are getting government payments because they are contracted to government and they are either not paying us… or they are not even on the register for us for certain products,” Makhubu said.
The flow of information is becoming more frequent too. Whereas SARS previously received procurement data monthly, it now receives updates every week, with the longer-term goal of moving to daily reporting.
That may sound like an operational detail, but it means SARS can identify potential compliance risks far sooner than it could in the past.
It’s not just about companies
The data matching doesn’t stop with businesses.
One example Makhubu gave illustrates just how far this capability has evolved.
SARS is also comparing company activity with the personal tax affairs of company directors.
“If their company is turning over R200 million just on government transactions alone, we expect that the directors should actually be paid quite well, and if we are not finding them at the right declaration level, the data is allowing us to be able to pick them up,” he said.
In other words, SARS is no longer looking at tax returns in isolation. It is increasingly building a broader picture by viewing company turnover, VAT registration, procurement payments, and personal tax declarations alongside one another.
That doesn’t represent a change in SARS’s powers. Rather, it reflects a significant improvement in how the revenue service uses information that is already available to it.
As its analytical capability expands, taxpayers who have previously managed to slip through the cracks may find there are fewer places left to hide.
Makhubu said SARS has already identified about 400 000 taxpayers who are outside the tax base but may need to be brought into the system as these data-matching capabilities continue to develop.
AI is the engine, not the story
AI featured prominently throughout the briefing, but Makhubu consistently spoke about it as a tool for improving decisions rather than replacing people.
One example is the way SARS now identifies fraudulent refund claims before money leaves the fiscus.
“We have stopped in excess of R100bn of impermissible refunds from going out by ensuring that our algorithms are able to read from various sources of data… and guide the decisions that are made,” he said.
Although he did not specify the period over which those refunds were prevented, the figure provides some indication of the scale at which SARS’s risk engines are now operating.
The technology isn’t replacing investigators. Instead, it helps SARS to distinguish far more quickly between taxpayers whose refunds can safely be paid and those that require closer scrutiny.
Better data, better service
The same technology that helps SARS to identify non-compliance is also improving the experience for compliant taxpayers.
SARS receives more than 21 million third-party data records from employers, banks, retirement funds, medical schemes, and other institutions. Those records are matched before tax returns are pre-populated and auto assessments are issued.
According to Carl Scholtz, SARS’s group executive for strategy enablement and modernisation, the better the data quality becomes, the more accurate those assessments become.
“The more accurate the data is, the more we can match, the more accurate the outcomes of the auto assessments, and the better the taxpayer experience,” he said.
AI is also behind Lwazi, SARS’s virtual assistant, and will increasingly be used to perform electronic verification of supporting documents and improve service through SARS’s contact centres.
A different kind of modernisation
Listening to Makhubu, it became clear that his vision for SARS is less about introducing flashy new technology and more about making the organisation work smarter.
Throughout the briefing he returned repeatedly to themes of better execution, better use of data, and removing unnecessary friction for compliant taxpayers, while making it increasingly difficult for those who choose not to comply.
“It is our conviction that, at the rate at which technology is moving, tax administrations that are not modernising at the same rate will certainly be left behind,” he said.
If Edward Kieswetter’s legacy was rebuilding SARS as an institution, Makhubu’s early focus appears to be on making that institution increasingly intelligent – not by replacing people with technology, but by giving people better information on which to make decisions.
That may ultimately prove to be one of the most significant changes to come out of this year’s filing season.
