What happened
The Office of the Tax Ombud reviews complaints about the South African Revenue Service (SARS). In 2017, with the Deputy Minister of Finance's approval, it investigated delays in refund payments. The Ombud analyzed SARS data for the 2017 financial year on refund credits that SARS identified for audit or verification. Audits and verifications cut those credits by 34,776,054,832 rand, or 17.38%. Of those cases, 504,468 ended in SARS's favor, 964,910 ended with no change and 31,818 ended in the taxpayer's favor. In a report dated 28 August 2017, Tax Ombud Judge B M Ngoepe concluded that SARS's risk engine seemed relatively effective in preventing losses. SARS told the Ombud that it identifies risks systematically, for example by comparing third-party data, and resolves them by manual review. The report does not describe how the engine scores claims or whether it learns from data. The Ombud also concluded that delayed refunds had become a systemic issue.
Sources
-
Audit report, by authors with no tie to the system.
Office of the Tax Ombud (South Africa), Judge B M Ngoepe. Tax Ombud's Report on the investigation in terms of Section 16(1)(b) of the Tax Administration Act 28 of 2011 into alleged delayed payment of refunds as a systemic and emerging issue. Dated 28 August 2017, Section V, paras 2.4.1 and 2.4.2.
This shows that a significant number of refunds claimed were correctly identified for audit/verification. It would therefore seem that SARS's risk engine is relatively effective in preventing losses through refund claims.
Details
AI system
- Workflow
-
SARS identified refund claims for audit or verification systematically, for example by comparing third-party data. SARS staff then resolved the identified risks by manual review.
Money
- Kind of value
-
Cut in the refund credits claimed on returns that SARS identified for audit or verification in the 2017 financial year, which the Ombud calls the amount that would have been paid out without those checks
- Amount
-
34,776,054,832 rand
- Recipients and payers
-
SARS, for the South African fiscus
Follow-up
- Conditions for replication
-
SARS must have enough audit capacity to finish reviews without holding back valid refunds. Refunds must be released promptly once a verification is done, and the Ombud found that SARS often failed to do so.
Entry details
- Sector
-
Tax administration.
- Purpose
-
The South African Revenue Service built its risk engine to identify refund claims for audit or verification before refunds are paid.
Not yet checked: model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, currency and prices, period and unit, realization, gross or net, included costs, comparison, method, return and payback, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). Tax Ombud finds that SARS's risk engine curbs refund losses (AIBD-0855). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0855.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)