SARS Auto-Assesses 5 Million Taxpayers, Paying Out R10 Billion in Refunds

SARS has been able to process over 90% of refunds for taxpayers not selected for verification within 72 hours using its fraud risk detection AI.


Devdiscourse News Desk | Pretoria | Updated: 16-07-2024 22:08 IST | Created: 16-07-2024 22:08 IST
SARS Auto-Assesses 5 Million Taxpayers, Paying Out R10 Billion in Refunds
SARS uses sophisticated graph database technology to analyze billions of data rows, allowing for a comprehensive view of each taxpayer. Image Credit: Twitter(@SAgovnews)
  • Country:
  • South Africa

The South African Revenue Service (SARS) has already disbursed approximately R10 billion in refunds to over 1.6 million taxpayers who were auto-assessed ahead of this year’s Tax Season. SARS Commissioner, Edward Kieswetter, shared this update during a media briefing on Tuesday.

In the first two weeks of July, SARS auto-assessed at least five million taxpayers. Kieswetter explained, “We completed these annual returns and tax assessments while, simultaneously, running each assessment outcome through our compliance risk and tax fraud detection capability. All this is possible because of the investment that we have been able to make in increasing and expanding the use of third-party data in the past few years.”

SARS has been able to process over 90% of refunds for taxpayers not selected for verification within 72 hours using its fraud risk detection AI. High-risk taxpayers, suspected of potentially fraudulent or impermissible refunds, are directed for further verification.

To ensure accurate auto-assessments, SARS harvested nearly 150 million third-party data records. This data was processed through artificial intelligence and enhanced machine learning algorithms. Kieswetter emphasized the importance of accurate and timely data from employers, banks, and other third-party information providers. “The integrity of the entire value chain of data is important. We have worked extremely hard this past year with large institutions and entities to improve the accuracy of bulk data submissions,” he said.

SARS uses sophisticated graph database technology to analyze billions of data rows, allowing for a comprehensive view of each taxpayer. This enables precise assessment outcomes and robust risk detection using AI methodologies. Kieswetter highlighted that the AI tool used by SARS has saved billions of Rands in fraudulent refunds. “Last year alone, this fraud detection capability prevented almost R100 billion of impermissible PIT and VAT [Personal Income Tax and Value Added Tax] refunds from being paid out and lost to the fiscus,” he noted.

SARS expressed gratitude to all third-party data providers for their partnerships, which have enabled a seamless experience for taxpayers. Inaccurate or late data submissions cause downstream difficulties and unnecessary delays for taxpayers.

Give Feedback