AI Generated Evidence in SMSF Audits

As AI Rises — So Must Professional Scepticism

The 2025/26 SMSF audit season has brought with it a new and rapidly emerging challenge.

At Saul SMSF we have notice a growing number of documents presented for audit, are no longer merely poorly prepared or incomplete.

They are AI generated.

• Not just trustee minutes.
• Not just investment strategies.
• Not just generic compliance paperwork.

We are now seeing AI-generated “history” seeking to fill gaps where evidence may never have existed at all:

• AI-generated deeds.
• AI-generated loan agreements.
• AI-generated explanations for gross overpayments of interest.
• AI-generated narratives attempting to defend serious Non-Arm’s Length Income (NALI) exposures.

This is no longer theoretical.

It is here.

And it is growing.

The concern is not the technology itself.

Artificial Intelligence is an extraordinary tool. Used properly, it can improve efficiency, enhance documentation processes and support better outcomes across the SMSF sector.

But AI must never become a mechanism to manufacture audit evidence after the fact.

This is the line in the sand.

Under Australian Auditing Standard ASA 500 Audit Evidence, the role of the independent auditor is clear “To obtain sufficient appropriate audit evidence to form an audit opinion”.

• Not assumed.
• Not reconstructed.
• Not synthetic.
• Not evidence created years later to “plug gaps”.

Real evidence.

• Evidence that is reliable.
• Evidence that existed.
• Evidence capable of supporting the substance of what actually occurred.

Because ultimately, the role of an SMSF auditor is not just “waving the audit through”.

Our role is to protect the integrity of the superannuation system.

And most importantly:

To protect member benefits.

Especially the weakest members.

• The silent spouse.
• The financially vulnerable member.
• The ageing retiree.
• The member with limited financial literacy.
• The member who trusts that the system around them is real, independent and functioning properly.

How many SMSF audits today are being waved through on weak foundations?

• How many SMSFs contain documents generated years after transactions occurred
• How many arrangements are being retrospectively papered over with AI?
• How many “commercial loan arrangements” never truly existed?
• How many Trustees now believe that if documentation is missing, technology can simply create it?

This is the real risk of AI – Not efficiency – But False legitimacy.

The emergence of AI-generated documentation should not reduce professional scepticism. It should dramatically increase it.

This is not merely a compliance risk.

It is people’s retirement savings.

Saul SMSF – Audits Well Solved.

David Saul