In a follow-up to our latest publication on the ATO’s use of artificial intelligence (AI) on an industrial scale, the ATO continues to scale up its efforts through AI to detect tax evasion and prevent taxation fraud from occurring.
The ATO recently identified more than $530 million of unpaid tax bills and stopped $2.5 billion from being fraudulently claimed with the assistance of artificial intelligence tools.
The ATO is using AI to trawl through its large data sets and deliver new insights that are impossible for humans to identify, which has emerged as a powerful tool in the agency’s fight to recover almost $45 billion in unpaid taxes owed by Australians.
Superannuation Underpayments
The ATO’s deep learning models have helped identify $295 million in superannuation guarantee underpayments.
Since the 2018 financial year, the ATO’s AI models which identify employers most likely to short-change staff on superannuation contribution entitlements, has achieved a 90% success rate in detecting superannuation underpayments, raising an additional $295 million in liabilities.
Busting GST Fraud
The ATO’s AI language models had scoured through leaked documents, such as the ‘Panama Papers’, to detect $242 million owed by tax evaders since 2018. This has resulted in the ATO collecting an additional $60 million in tax and finalised 535 audits and reviews.
The ATO is also using ‘gradient-boosting’ machine learning models to identify evolving fraudulent GST behaviour. Gradient-boosting models work by building models in sequence, where each successive model tries to reduce the errors in the previous model.
As at 31 December 2022, the ATO had taken compliance action against more than 53,000 taxpayers and stopped about $2.5 billion in fraudulent GST refunds from being paid to individuals and businesses seeking to defraud the GST system.
The ATO has stated that it is “using AI to amplify the capacity of its staff by developing tools that assemble relevant information and undertake the ground-level pattern recognition to free staff to concentrate on the tasks requiring human judgment and empathy”, in the fight against tax evasion.