← Field Notes

AI Compliance & Regulation

AI-Forged Payslips: What Australian Mortgage Brokers Must Verify

AUSTRAC referred about 200 brokers, accountants and lawyers in its mortgage fraud probe, and AI is making fake payslips. What ASIC expects brokers to verify.

In short. A payslip on its own is no longer enough to verify income. ASIC's responsible lending guide (RG 209) says brokers carry their own duty to verify, should not rely on information they have reason to doubt, and can confirm a payslip against bank transaction data or the employer. With AUSTRAC reporting AI-made fake documents in its mortgage fraud investigations, check income at the source, not the PDF.

Does software help decide things about your customers or staff? From 10 December 2026 your privacy policy may have to say so. Six questions give a likely answer.

Take the check →

About 200 brokers, accountants and lawyers were referred. Fake documents were the common thread

AUSTRAC's mortgage fraud investigation, Operation Claw, analysed data from 10 major banks, found potentially hundreds of millions of dollars in suspect home loans, and referred around 200 mortgage brokers, lawyers, accountants and companies to police, tax authorities and the corporate regulator. The warning signs it found again and again were falsified or misleading documents, and the same brokers, accountants and law firms appearing across many applications. This week AUSTRAC's chief executive said the agency has already seen criminals use AI in those investigations, mainly for synthetic identities and fake documents.

If your business puts a client's income documents in front of a lender, the question is no longer whether a payslip could be fake. It's whether your process would catch one.

~200

Brokers, lawyers, accountants and companies referred

AUSTRAC, Operation Claw

10

Major banks' data analysed

Fintel Alliance

~1,800

Suspicious matter reports after the first alert

As reported by The Adviser

What happened

Operation Claw started in February 2026, after the Commonwealth Bank flagged concerning patterns in its loan book, and AUSTRAC made the findings public in August. As reported by The Adviser, suspect applications featured inflated incomes, misrepresented employment and business activity that was made up or couldn't be checked. AUSTRAC's message to lenders: "The most effective way to stop mortgage fraud is before a loan is approved."

At ASIC's Annual Forum in Sydney this week, chair Sarah Court said AI is "making false material more convincing and scams easier to execute at scale", and then added the line that matters for every licensee: "AI does not change firms' accountabilities or obligations." The same week, ASIC named poorly run AI and automated systems among its 2027 enforcement priorities.

What your verification duty already says

None of this needs a new law. ASIC's Regulatory Guide 209 on responsible lending already covers it, and four passages do most of the work.

First, the broker's duty is your own. RG 209.129 says a broker must make inquiries and verify information to meet its own obligations, "not merely to support the lender's ability to make a final assessment." The lender's checks don't cover you. Second, the guide expects fraud. RG 209.48 says verification exists because information may be unreliable, including through "deliberate fraud by the consumer or a person who is assisting the consumer."

Third, a payslip alone can fail the test. The guide cites the Federal Court's 2018 ruling against ANZ, which failed to take reasonable steps when it relied only on a purported payslip it knew was a type of document "easily falsified". When there is reason to doubt a document, RG 209.133 suggests checking it against a separate source, such as a transaction statement, the employer, or a digital data capture service. Fourth, the standard moves. RG 209.123 says what counts as reasonable "will change over time" as better ways to verify become available.

Put those together and the direction is clear. When AI makes a convincing fake payslip cheap to produce, and bank transaction data can be pulled with the client's consent, relying on a PDF gets harder to defend each year.

AI for mortgage brokers in Australia: use it to check, not just to process

The tools that forge documents also check them. Lenders are already responding: in May, The Adviser reported that non-bank lender RedZed had added automated document and image fraud detection to its application process, flagging mismatches as documents arrive. A brokerage can do the same on its side. That means income confirmed from bank data rather than an uploaded PDF, automatic checks for edited files and mismatched metadata, and alerts when the same accountant or referrer keeps appearing on files.

It is also an admin leverage gain, not just a compliance cost. Pulling income from source data takes less time than chasing, reading and re-keying PDFs, and it leaves a record of how you verified each file. That record is what you will want if a file is ever questioned. If you'd like to see where AI checks would fit into your current fact-find and lodgement steps, book a short call and we'll go through it with you.

Accountants carry a version of the same risk. An income letter for a self-employed client is exactly the document AUSTRAC saw repeated across suspect files, and since 1 July many accounting and legal practices are AML/CTF reporting entities in their own right.

What to do this month

Pull your last 20 settled files and note how income was verified on each: PDF only, PDF plus bank data, or source data. Write a rule for when a payslip needs a second source, and make it the default for self-employed clients, new referrers and files where the numbers don't match. Then look at where AI can do the checking for you, inside the software you already use. Our page on AI for mortgage brokers sets out how we approach it, and our earlier note covers AI and the best interests duty.

Key takeaways

01AUSTRAC's Operation Claw referred around 200 mortgage brokers, lawyers, accountants and companies; falsified documents and repeat intermediaries were recurring warning signs.
02AUSTRAC says criminals have used AI for synthetic identities and fake documents in the mortgage fraud investigations; ASIC says AI does not change firms' obligations.
03Under RG 209 a broker's verification duty is its own, and relying only on a payslip can fall short, as in ASIC v ANZ (2018).
04Verify income against source data such as bank transactions or the employer, and use AI tools to flag edited documents and repeat referrers.

Common questions

Is a payslip enough to verify income for a home loan in Australia?

Not always. ASIC's RG 209 cites the 2018 Federal Court case against ANZ, which failed to take reasonable steps by relying only on a purported payslip. Where there is reason to doubt a document, ASIC suggests checking it against another source, such as transaction statements or the employer.

Can a mortgage broker rely on the lender to verify a client's income?

No. RG 209.129 says a broker's inquiries and verification are to meet its own obligations, not merely to support the lender's final assessment.

Are criminals using AI to commit mortgage fraud in Australia?

AUSTRAC's chief executive said in November 2026 that the agency has seen AI used in its mortgage fraud investigations, mainly to create synthetic identities and fake documents.

Sources

Assumptions & methodology
  1. Operation Claw details (start in February 2026, data from 10 major banks, potentially hundreds of millions of dollars in suspect loans largely linked to Sydney properties, around 200 referrals, around 1,800 suspicious matter reports, and the quoted AUSTRAC lines) are as reported by The Adviser (19 August 2026), drawing on AUSTRAC's Fintel Alliance release. AUSTRAC's own site was unavailable when we checked it, so we cite the report.
  2. Sarah Court's quotes are from ASIC's published opening remarks to the ASIC Annual Forum 2026. Brendan Thomas's comments on AI use in the mortgage fraud investigations are as reported by The Adviser (20 November 2026). The RedZed example is from The Adviser (18 May 2026).
  3. RG 209 references are to the December 2019 version of ASIC Regulatory Guide 209, including the note to RG 209.132 on ASIC v Australia and New Zealand Banking Group Limited [2018] FCA 155. The 20-file review is a CoterieLabs suggestion, not a regulatory requirement. Last reviewed 22 November 2026 · general information, not legal advice.
ShareLinkedInEmail

Field Notes are general commentary on AI trends for Australian businesses. They don’t constitute professional advice. Talk to your accountant, lawyer, or IT adviser before acting on anything specific to your situation.

If this applies to you

Would your files hold up if a payslip turned out to be fake?

We look at how your brokerage verifies income today, show where AI checks can catch edited documents and repeat referrers, and set them up inside the tools you already use. Book a call to start with your own files.

Don't miss the next one

Get each new Field Note in your inbox as it publishes — short, practical AI intelligence for business owners.