SARS Modernizes with Data-Driven Approach

The South African Revenue Service (SARS) has quietly become one of Africa’s most sophisticated users of artificial intelligence, transforming tax administration across the board.

What was once a manual process requiring extensive document review is now largely automated through data analytics, machine learning and AI. These technologies assess risk profiles, match third-party information and flag returns for closer scrutiny—often before taxpayers even submit them electronically.

Impact on Tax Compliance

According to SARS Head of Communication Siphithi Sibeko, the compliance program contributed R304 billion ($18.2 billion) in the 2024/25 financial year. AI-powered fraud detection prevented over R417 billion ($25 billion) in improper refunds during the past five years.

Key highlights:

• Over 100% of verification cases now use automated risk assessment
• 88.41% of complex audits selected through algorithms
• More than 1.9 million taxpayers have been auto-assessed, with R8 billion ($479 million) in refunds processed within 72 hours

Expanding Data Ecosystem

SARS built a comprehensive digital view by integrating data from employers, banks, insurers, investment firms and even foreign tax authorities.

This is particularly crucial as South Africa’s digital economy grows—online retail sales are projected to reach R130 billion ($7.8 billion), or nearly 10% of total retail.

“The focus isn’t on any single technology, but a platform approach that integrates data, analytics and AI to make compliance easier for honest taxpayers while deterring those who evade,” Sibeko explained.

Human Oversight Remains Critical

While AI handles much of the initial screening, SARS maintains human oversight throughout the process. Risk indicators are reviewed through governance frameworks before any enforcement action is taken.

This “human-in-the-loop” approach addresses growing concerns about algorithmic accountability and transparency in public decision-making.