Banks Push for ATO Data as AI Mortgage Fraud Escalates Across Australian Lending
Major banks have told a Senate committee that AI generated fake documents are making mortgage fraud easier to commit and harder to detect. Their proposed solution is direct, secure access to Australian Taxation Office data so borrower income can be verified at the source rather than through payslips, tax returns and bank statements supplied as PDFs.
The warning marks another escalation in Australia’s mortgage fraud crisis.
On 21 July 2026, Mortgage Professional Australia reported that Westpac’s chief economist, Luci Ellis, described the problem as “burgeoning” and likely to worsen as more fraudsters understand what generative AI tools can produce (Mortgage Professional Australia, 2026). Australian Banking Association chief executive Simon Birmingham told the Senate committee that it was concerning that sensitive documents such as payslips, tax returns and bank statements were still routinely gathered and submitted by hand, or by email (Australian Banking Association, 2026).
This is not simply a debate about open banking. It is a direct challenge to Australia’s document based lending model.
For decades, banks have relied on documents to verify borrower income, assess serviceability and approve loans. That model assumes documents are reliable inputs. In 2026, that assumption is under pressure.
AI can now generate realistic payslips. It can assist in doctoring bank statements. It can support the creation of false tax material, synthetic business records and internally consistent loan files. What once required technical skill can now be produced faster, cheaper and with greater visual credibility.
The question for lenders is no longer whether fraud exists in mortgage origination. The question is whether existing verification systems can detect it before funds are released.
Why Banks Want ATO Data Access
The banking industry’s push is straightforward. Instead of relying on borrower supplied documents, lenders want consent based access to verified income data held by the Australian Taxation Office.
The proposed mechanism is the Consumer Data Right, Australia’s open data sharing framework. The CDR went live for banking data sharing in July 2020 and was designed to give consumers greater control over how their data is shared with accredited recipients (Consumer Data Right, 2020). Banks now argue that the same framework could be expanded to include ATO income data, provided legislative barriers such as Taxation Administration Act secrecy provisions are addressed.
The reason is clear.
If a borrower provides a payslip, the bank must decide whether that document is genuine. If the bank can verify income directly against ATO records, the opportunity for forged or AI generated income documents reduces significantly.
This would not remove the need for lending judgement. Nor would it eliminate fraud entirely. However, it would shift verification closer to a source of truth.
That is the critical distinction.
Document based verification asks whether a submitted document looks right. Source based verification asks whether the underlying claim is true.
The Mortgage Fraud Scandal Has Moved Beyond One Bank
The call for ATO data access follows months of escalating mortgage fraud revelations.
In February 2026, Commonwealth Bank reportedly self reported suspected fraudulent home loans to police after concerns emerged that applications had been supported by forged or AI generated documents. What began as an estimated $1 billion exposure was later reported by Mortgage Professional Australia as closer to $4 billion across the five biggest lenders, with all ten of Australia’s largest banks auditing their books (Mortgage Professional Australia, 2026).
National Australia Bank confirmed in June that it had taken action against multiple parties involved in mortgage fraud, including referring matters to authorities and exiting or suspending individuals and entities from the bank. NAB described the issue as an industry wide threat requiring stronger collaboration between banks, regulators, law enforcement and government (NAB, 2026).
At the same time, ASIC confirmed that it was working with AUSTRAC, state police and the major banks to address syndicated mortgage fraud moving through the broker channel. ASIC commissioner Alan Kirkland warned that coordinated conduct across multiple parties, sometimes linked to broader criminal enterprises, was a serious and evolving threat to confidence in home lending (Mortgage Professional Australia, 2026).
This is why the Senate evidence matters.
The industry is no longer discussing fraud as a narrow operational issue. It is now being framed as a systemic verification problem across banks, brokers, accountants, referrers and professional facilitators.
AI Generated Documents Expose the Weakness in Manual Verification
The mortgage sector’s reliance on PDFs, scanned files and manually submitted evidence has become a structural vulnerability.
A payslip can be edited. A bank statement can be manipulated. A tax return can be drafted but not lodged. A company record can be created to support a false trading history. A borrower profile can be assembled to appear serviceable across multiple documents.
When generative AI enters this environment, the threat escalates.
Traditional manual verification was built for visible errors. A reviewer could inspect fonts, layout, numbers and signatures. Basic OCR could extract text and compare fields. A checklist could confirm whether required documents were present.
However, modern fraud often appears complete and it may look consistent. It may align across a loan file. It may not reveal itself through obvious visual defects.
That is the new risk.
Fraudsters are no longer only forging documents. They are engineering credibility.
ATO Data Helps, But It Does Not Solve the Full Problem
Secure access to ATO income data would be an important reform. It would help lenders verify income at the source and reduce reliance on borrower supplied payslips and tax documents.
However, it would not remove the need for forensic document fraud detection.
Mortgage applications rely on more than income alone. Banks still receive bank statements, identity documents, company records, trust structures, contracts, invoices, living expense declarations, asset documents and supporting evidence from brokers, accountants and customers.
Even if income can be verified through the ATO, fraud can still enter the process through manipulated documents, shell entities, altered account details, synthetic identities or inconsistent supporting files.
This is why the future of mortgage verification cannot rely on one control.
It requires a layered model.
Source data can verify whether declared income is accurate. Fraud detection AI can determine whether submitted documents have been manipulated. Cross document validation can assess whether the entire application makes sense.
That combination is where the industry is heading.
Fraud Check AI and the Shift From Document Review to AI Document Forensics
Fraud Check AI, developed by DoxAI, is designed for the exact problem now confronting lenders.
It does not treat a document as trustworthy because it looks professional. It examines whether the file has been manipulated, generated, altered or structured in a suspicious way.
The first layer is static analysis. This examines file structure, metadata, encoding signatures, embedded objects, hidden artefacts and digital fingerprints. These indicators can reveal manipulation that is not visible to a human reviewer.
The second layer is dynamic validation. This assesses whether the information inside the document is logical. In a lending context, income figures, dates, pay periods, tax amounts, balances and calculations must make sense. A document can look genuine but still fail basic logic when examined properly.
The third layer is custom risk control. A bank may want to flag unusual overseas deposits, repeated broker patterns, high risk company structures, unusual income movements or applications involving specific document combinations. Fraud Check AI allows these controls to align with institutional risk appetite.
This is the shift banks must make.
The industry cannot rely only on people checking what appears on a page. It must analyse how a document was created and whether it should be trusted before the loan progresses.
Why Cross Check AI Becomes Critical When ATO Data Expands
ATO data access would strengthen income verification, but income is only one part of the lending decision.
Cross Check AI helps validate whether information is consistent across documents, systems and business rules.
- If a borrower’s declared income is verified by ATO data, Cross Check AI can compare that against bank statement deposits, employment information, company records and declared liabilities.
- If an application includes business income, it can compare BAS statements, financial statements and account activity.
- If broker introduced documents appear consistent individually, Cross Check AI can examine whether the broader application tells a coherent story.
This matters because modern mortgage fraud is rarely built on one fake file. It is built across multiple documents.
A payslip supports a salary. A bank statement supports deposits. A tax return supports income history. A company file supports trading activity. An identity document supports the applicant profile.
- Fraud Check AI asks whether the documents have been manipulated.
- Cross Check AI asks whether the information across the workflow can be trusted.
Together, they create a stronger verification model than manual review alone.
CEO Perspective
Alfonso Porcelli, Chief Executive Officer of DoxAI, says the Senate push for ATO data access reflects a broader shift in how lenders must think about verification.
“Banks are asking for ATO data because they recognise that document based verification is no longer sufficient on its own. AI generated documents can now appear complete, consistent and credible. The challenge is not only verifying income at the source, but also determining whether every supporting document in the workflow can be trusted.”
He continues.
“The future of mortgage fraud prevention will be layered. Source data, forensic document analysis and cross document validation must work together. Fraud Check AI was built to detect document manipulation before approval, while Cross Check AI validates whether information remains consistent across the entire application.”
What This Means for Australian Banks and Brokers
For banks, the message is direct. Mortgage verification must move upstream. Fraud cannot be detected only after settlement, whistleblower complaints or portfolio reviews. It must be identified inside origination workflows before approval.
For brokers, the scrutiny will intensify. Senator Andrew Bragg reportedly noted that more than three quarters of home lending flows through broker networks, making broker adoption central to any reform (Mortgage Professional Australia, 2026). If banks gain access to ATO data, broker submitted income material may become easier to verify. However, brokers will also need stronger systems, better training and clearer accountability across document collection and submission.
For regulators, the issue is now broader than lending misconduct. It touches AML, professional facilitation, open data policy, privacy, cyber security and consumer protection.
For borrowers, the benefit could be faster verification and fewer document requests. However, this will only work if privacy, consent and governance remain central to the model.
The Future of Mortgage Fraud Prevention Is Source Verified and AI Governed
The banks’ request for ATO data access is a signal that the industry is moving away from blind trust in documents.
That does not mean documents disappear. It means documents must be treated as evidence to be tested, not assumptions to be accepted.
In the next phase of mortgage origination, high trust lending will depend on three capabilities.
- The first is source verification, where income and identity claims are checked against trusted datasets with customer consent.
- The second is forensic document analysis, where AI detects manipulation, synthetic creation and hidden file anomalies.
- The third is cross document validation, where information is checked across the full application before decisions are made.
This is how lenders can reduce exposure to AI generated fraud while improving approval speed for genuine borrowers.
The industry has reached a turning point. Fraud now moves faster than manual controls. Verification must move faster too.
If your organisation is reassessing mortgage, broker or income verification controls, now is the time to strengthen your defences before the next wave of AI generated fraud enters the loan book.
Explore how DoxAI’s Fraud Check AI and Cross Check AI can help detect manipulated documents, validate borrower information and strengthen lending workflows before risk becomes loss.
About DoxAI
DoxAI is the automation partner for enterprises ready to become AI-native. We replace outdated systems and fragmented workflows with one intelligent automation layer unifying the collection, management, processing, and storage of data and documents to deliver stronger security, lower costs, and better customer experiences at scale.
DoxAI combines AI Governance, 70+ automation solutions, 90+ AI agents, Alfy AI and proprietary AI models. Its architecture also supports integration, consulting and sovereign AI deployment.
The platform supports financial services, government, healthcare, insurance, legal and other regulated industries. DoxAI can integrate with existing enterprise systems through APIs and configurable workflows.
References
Australian Banking Association (2026) ABA opening statement to Senate Select Committee on Productivity in Australia. Available at: https://www.ausbanking.org.au/aba-opening-statement-to-senate-select-committee-on-productivity-in-australia/ (Accessed: 29 July 2026).
Consumer Data Right (2020) Consumer Data Right goes live for data sharing. Available at: https://www.cdr.gov.au/news/media-releases/consumer-data-right-goes-live-data-sharing (Accessed: 29 July 2026).
Mortgage Professional Australia (2026) Give us access to tax data to fight mortgage fraud, major banks tell senate. Available at: https://www.mpamag.com/au/news/general/give-us-access-to-tax-data-to-fight-mortgage-fraud-major-banks-tell-senate/583002 (Accessed: 29 July 2026).
Mortgage Professional Australia (2026) ASIC addresses brokers on mortgage fraud: With power comes responsibility. Available at: https://www.mpamag.com/au/news/general/asic-addresses-brokers-on-mortgage-fraud-with-power-comes-responsibility/583229 (Accessed: 29 July 2026).
NAB (2026) NAB takes action on industry-wide mortgage fraud. Available at: https://www.publicnow.com/view/D90BB4B08F8710CA6A815B15B6895B5EBB77FDFB (Accessed: 29 July 2026).
Author
