Algorithmic Arm's-Length—AI-Based Risk Scoring and Procedural Fairness in Transfer-Pricing Audits
Canadian Tax Journal/Revue fiscale canadienne · 2026-07 · 74 (2) · 419-433
DOI: 10.32721/ctj.2026.74.2.pf.narayanan ↗Kurz gefasst
Fachartikel in der Zeitschrift Canadian Tax Journal/Revue fiscale canadienne, 2026-07. Verfasst von Theertha Narayanan und 1 weiteren Autor:innen.
Abstract (Originalsprache)
This article examines how the use of artificial intelligence (AI)-based risk scoring for transfer-pricing audits by the Canada Revenue Agency (CRA) interacts with Canadian administrative law and emerging international AI governance. The article argues that the transfer-pricing reforms introduced in the 2025 federal budget—stricter methodologies, expanded recharacterization powers, higher penalty thresholds, and shortened documentation deadlines—heighten the duty of procedural fairness in audit selection. The article reviews doctrinal standards under Vavilov and Dow Chemical, compares CRA practice with OECD guidance and foreign developments, and assesses whether algorithmic selection can remain justified, intelligible, and transparent. It concludes by proposing a governance code for AI-based tax audits to reconcile efficiency with legality and taxpayer rights.
Zitieren
APA
Narayanan, T., & Siva, P. K. (2026). Algorithmic Arm's-Length—AI-Based Risk Scoring and Procedural Fairness in Transfer-Pricing Audits. Canadian Tax Journal/Revue fiscale canadienne, 74(2), 419-433. https://doi.org/10.32721/ctj.2026.74.2.pf.narayanan
Vancouver
Narayanan T, Siva PK. Algorithmic Arm's-Length—AI-Based Risk Scoring and Procedural Fairness in Transfer-Pricing Audits. Canadian Tax Journal/Revue fiscale canadienne. 2026;74(2):419-433. doi:10.32721/ctj.2026.74.2.pf.narayanan
BibTeX
@article{Narayanan2026,
author = {Narayanan, Theertha and Siva, Pramod Kumar},
title = {Algorithmic Arm's-Length—AI-Based Risk Scoring and Procedural Fairness in Transfer-Pricing Audits},
journal = {Canadian Tax Journal/Revue fiscale canadienne},
year = {2026},
volume = {74},
number = {2},
pages = {419--433},
doi = {10.32721/ctj.2026.74.2.pf.narayanan}
}
EUVIMED