Can anyone share some practical recommendations as to what specific activities Internal Audit can perform over AI models, specifically in the areas of bias, transparency, model drift, model validation...etc.?

128 viewscircle icon2 Comments
Sort by:
Chief Financial Officer7 hours ago

In my view, the Internal Audit dep. is actually one of the functions that can benefit the most from AI.
AI can help automate and standardize if not all, Most of the control measures based on existing SOPs and policies, ensuring compliance, reducing human error and speeding the proceed. Also with the right algorithm, AI can detect anomalies and possible fraud through specific, data-driven commands.
For example, in a retail environment, I used to perform sensitivity analyses comparing refund patterns versus total transactions, tracking Percentage of what was refunded, when, and by whom, ect…. Those reports helped us uncover several fraud cases of fake refunds. However, I believe with AI, this process could be way faster, more dynamic, and can make fraud detection far more efficient and proactive.
So while internal audit teams should indeed perform oversight on AI models for bias, transparency, or drift, they can also leverage AI as a partner to strengthen assurance, risk monitoring, and fraud analytics.

Chief Financial Officera day ago

In my view, the Internal Audit dep. is actually one of the functions that can benefit the most from AI.
AI can help automate and standardize if not all, Most of the control measures based on existing SOPs and policies, ensuring compliance, reducing human error and speeding the proceed. Also with the right algorithm, AI can detect anomalies and possible fraud through specific, data-driven commands.

For example, in a retail environment, I used to perform sensitivity analyses comparing refund patterns versus total transactions, tracking Percentage of what was refunded, when, and by whom, ect…. Those reports helped us uncover several fraud cases of fake refunds. However, I believe with AI, this process could be way faster, more dynamic, and can make fraud detection far more efficient and proactive.

So while internal audit teams should indeed perform oversight on AI models for bias, transparency, or drift, they can also leverage AI as a partner to strengthen assurance, risk monitoring, and fraud analytics.

Content you might like

Sales11%

Marketing27%

Accounts Payable22%

Accounts Receivable13%

Legal6%

HR16%

Other (pls comment)2%

View Results

Coverage—AI claims full scan, but misses deep flaws49%

Speed—AI is fast but error-prone63%

Creativity—AI scripts can’t improvise10%

Integration—vendor tools don’t plug into DevSecOps27%

View Results