TLDR: Edward Kieswetter, Commissioner of the South African Revenue Service (SARS), announced at the G20 Zimbali meeting that artificial intelligence (AI) and machine learning are significantly boosting SARS’s efficiency in tax collection. He revealed that AI has already assessed 65% of taxpayers without requiring them to submit returns, leading to rapid refunds and assessments, and preventing billions in impermissible claims. This showcases AI’s practical application in government services, while also prompting discussions on balancing innovation with responsible regulation.
Edward Kieswetter, the Commissioner of the South African Revenue Service (SARS), recently highlighted the transformative impact of artificial intelligence (AI) on tax administration during a G20 Finance Track side event at the Zimbali Resort in KwaZulu-Natal. Kieswetter stated that AI and machine learning are pivotal in enhancing SARS’s operational efficiencies and improving tax collection processes.
According to Kieswetter, SARS has leveraged AI and machine learning to assess a remarkable 65% of taxpayers within the first nine days of the current tax filing season, eliminating the need for these individuals to submit a tax return. This achievement is based solely on third-party data, with AI algorithms performing assessments and sophisticated fraud risk detection. This marks a significant leap from two decades ago, when it took SARS six months to assess six million taxpayers. Now, for the two million taxpayers who opt to file, assessments are completed in under five seconds.
The practical benefits of this AI integration are substantial. Within the initial nine days of the 2025 tax season, SARS successfully paid out R12 billion in refunds and raised R3.5 billion in assessments. Furthermore, the strategic deployment of AI has been instrumental in preventing R5.5 billion in impermissible refunds, underscoring its critical role in safeguarding public funds.
Kieswetter emphasized the importance of a balanced approach to technological adoption, cautioning that ‘If policy gets too far ahead of practice, we may in fact suppress the experimentation and the exploration of new technology.’ This perspective aligns with broader discussions at the G20, which acknowledges the ‘profound duality’ of AI – presenting immense opportunities alongside significant risks.
Lesetja Kganyago, Governor of the South African Reserve Bank, echoed this sentiment, noting that while AI can enhance risk detection, improve consumer credit scoring, and deepen financial inclusion, it also carries the potential to amplify market shocks and perpetuate biases. Regulators are urged to adapt to these new technologies while remaining committed to their core mandates.
Also Read:
- KPMG’s Latest Global Report Highlights Accelerating AI Integration Across Finance Functions
- HMRC Pioneers Secure Generative AI Adoption with Dedicated ‘Landing Zone’
The G20 South Africa Presidency has prioritized ‘Equitable, Inclusive, and Just Artificial Intelligence (AI)’ as a key focus area. Discussions at the summit also included workshops on generative AI, its ability to produce deep fakes, and the impact on information integrity, aiming to identify strategies and regulatory frameworks to guide the responsible adoption of AI globally.
