AI for Financial Services Firms

AI pays first in a financial services firm on the preparation around advice and lending: meeting notes, client review packs, document collection, marketing compliance checks and complaints triage. Personal advice, credit decisions and anything that tells a customer no stay with licensed people. ASIC is clear that existing obligations apply to AI, including in advertising.

This is general information, not legal or financial advice.

Where does AI pay first in a financial services firm?

It pays first on preparation that a licensed person can check in minutes. That means file notes from client meetings, review packs assembled from existing records, chasing and sorting client documents, checking marketing drafts against approved wording, and flagging possible complaints. None of it is advice, and all of it frees advisers to give more.

Financial services is ahead of most sectors here. According to BCG's 2025 AI at Work survey, half of companies are moving beyond productivity plays to redesign whole workflows. Financial services and technology lead that shift. So the question for a smaller firm is less whether to start than where.

What should stay with licensed people?

Keep personal advice and its sign-off, credit and responsible lending assessments, investment recommendations, hardship and vulnerability decisions, and any communication that declines or changes what a customer receives. AI can gather the facts and draft. A licensed, accountable person decides, because the licence obligations sit with them and the firm.

ASIC's own review found this is how most licensees already use AI. Its 2024 report looked at 23 licensees across banking, credit, insurance and financial advice, and 624 AI use cases. ASIC's chair noted that use "has to date focussed predominantly on supporting human decisions". He also said "Existing consumer protection provisions, director duties and licensee obligations put the onus on institutions".

The same review found governance lagging. Close to half of the licensees lacked policies on consumer fairness or bias. Fewer still had policies on telling consumers when AI was used. A small firm can fix both in a page.

What does ASIC say about AI in financial advertising?

ASIC's updated advertising guide says the law applies to AI-generated advertising exactly as it does to content written by people, and that AI's tendency to hallucinate raises the risk of misleading ads. Advertising for services must also give a balanced view of any customer tools used, including AI tools.

According to RG 234, published in June 2026, "the laws relating to financial and credit products and services apply in the same way to AI-generated advertising content as human-generated content". The guide adds that "the potential for hallucinated or biased content by AI tools may increase the risk of misleading or deceptive advertising."

It also covers firms that promote AI as a feature. Advertisements for financial and credit services "should also give balanced messages about features, benefits, risks and limitations of those services and any customer tools used as part of those services, including AI tools." A firm advertising an AI-assisted service must describe its limits as clearly as its benefits.

In practice, every marketing draft should read from one approved list of product facts and wording, and compliance should check it before release. The guide to approved claims shows how to build that list.

How should client data be handled?

Client financial and personal information should never go into public AI tools. Use business plans with appropriate contract terms, connect AI to client systems read only, and keep each client's data in the firm's own accounts. APRA-regulated entities must also manage the operational and data risks of any AI service they rely on.

The OAIC recommends, as a matter of best practice, that organisations do not enter personal information, and particularly sensitive information, into publicly available generative AI tools. Staff using personal AI accounts with client data should move onto business plans first. The guide to whether Claude is safe explains how business plans handle data.

For banks, insurers and super funds, APRA's CPS 230 has applied since 1 July 2025. It sets minimum standards for managing operational risk, including service provider management. Its data risk guide, CPG 235, adds that data validation should "occur as close to the point of capture as possible", which is a useful rule for any AI workflow.

How should a financial services firm start?

Start with file notes or review pack preparation, measured against how long they take today. Move staff onto business AI plans, connect AI to the client system read only, have a licensed person check every output for a month, and record who approved each one. Then measure before choosing the next workflow.

1. Business plans first. Client data in personal accounts is a risk the firm cannot see.
2. One workflow. File notes are a good start, because the adviser was in the meeting.
3. Read only. AI reads client records; it changes nothing.
4. Recorded approval. Note who approved each output and when.
5. Measurement. Track time per review and how much each draft needed changing.

Write the approval points down before the first workflow runs. The guide to approval gates explains how.

Why does a financial services firm need shared knowledge?

Because product details, disclosures, approved wording and the firm's advice philosophy change, and each adviser's AI otherwise works from its own outdated version. One current, owned source of what is true about the firm and its products keeps every file note, review pack and marketing draft consistent and checkable.

That source holds approved product facts and wording with an owner for each, the firm's standard processes, and the reasons behind past decisions. When something new contradicts an existing entry, it goes to a person rather than being merged. Each client's own information stays in the firm's client systems, never in shared method files.

Skein builds a Company OS for financial services firms: what is true about the business held in sets that are each canonical, sourced and owned, with every connection starting read only and nothing published, sent or spent without a named person approving it; the Company OS page shows the three layers. That suits a firm of 10 to 200 people whose staff already use AI on their own. A large institution with its own AI platform team, or a firm wanting AI to give advice unchecked, is not a fit.

Frequently asked questions

How is AI used in financial services?

Smaller firms mostly use it for preparation: file notes from client meetings, review packs, document collection, checking marketing drafts against approved wording, and flagging possible complaints. A licensed person checks each output. Personal advice, credit decisions and anything that changes what a customer receives stay with accountable people.

Can financial advisers use AI to write statements of advice?

AI can help assemble facts and draft sections from the firm's own templates and the adviser's notes. The advice itself, and the decision to sign it, must stay with the licensed adviser, who remains responsible for its accuracy and appropriateness. Every figure and recommendation should be checked against client records before it goes out.

Does ASIC regulate AI in financial services?

ASIC has not written a separate AI rulebook. Existing consumer protection laws, director duties and licensee obligations apply to AI use. Its June 2026 advertising guide states that the same laws apply to AI-generated advertising as to human-written content, and that AI tools offered to customers must be advertised with their limitations.

Can I put client information into ChatGPT?

Not into a public AI tool. The OAIC recommends organisations avoid entering personal information, particularly sensitive information, into publicly available generative AI tools. Use a business plan with suitable terms, connect AI to client systems read only, and keep each client's data in the firm's own accounts.

Where should a small financial services firm start with AI?

Start with file notes or review pack preparation. Move staff onto business AI plans, connect AI to the client system read only, have a licensed person check every output for a month and record who approved it. Measure time saved and the edits each draft needed before choosing the next workflow.

Related reading