What should a compliance officer ask before the firm approves AI?
Ten questions for compliance officers and principal brokers deciding how their people may use AI, from which accounts are allowed to what the record will show an examiner.
Before approving AI, a compliance officer should know which AI accounts are allowed, whether client data trains the vendor’s models, what AI may never be used for, how client-facing output is checked, who reviews held drafts, how people accept the policy, how a tool is blocked, where data is stored, and what record an examiner will see.
The ten questions
- Which AI accounts are allowed? Personal free accounts, business accounts, or only accounts the firm provisions?
- Does our data train the vendor’s models? Business and enterprise plans from the major vendors say they don’t by default; confirm for each approved tool.
- What is AI never used for? For example credit decisions, final advice, or anything sent without a person reading it.
- What client data may never be entered? Identity numbers, account numbers, full financial details.
- How is client-facing output checked? Against which rules, and before or after it’s sent?
- Who reviews what gets held? Anti-money-laundering concerns, payment-detail changes and complaints need a named person.
- How do people accept the policy? Recorded, dated, against a version.
- How is an unapproved tool blocked? Refused at connection, or only discouraged?
- Where is the data stored and processed? And what does the vendor keep?
- What will the record show an examiner? What was checked, caught and accepted, without exposing client conversations.
How Broker AI answers them
Approved tools are checked on every request and unapproved ones refused and logged. The policy is versioned and every acceptance recorded. Client-facing drafts are checked against the firm’s rules, with financial-crime concerns held for a person. Data is stored and processed in Canada and no client data is stored. See it for firms.
General information, not legal or compliance advice. Rules differ by state, province and regulator; check with your compliance team or counsel.
Questions people ask
- Who is responsible for AI use at a brokerage?
- Typically the person already responsible for supervision, such as the principal broker or compliance officer. AI doesn’t create a new duty; it adds a new way for existing duties to be missed.
- Do we need a separate AI policy?
- Most firms add an AI section to existing policies or write a short standalone policy. What matters is that it names approved tools and forbidden uses, and that acceptance is recorded.
- Is this legal advice?
- General information, not legal or compliance advice. Rules differ by state, province and regulator; check with your compliance team or counsel.
By DisruptionIO, the team behind Broker AI. Published October 11, 2026.