AI certifications for banking: ABA courses, GARP RAI and the 2026 model risk rules

Banking has no single standard AI certification. The American Bankers Association runs two four-week AI courses, one for all staff and one for risk and compliance teams. GARP offers the Risk and AI (RAI) Certificate, which ends in a four-hour exam. The guidance on bank models changed in April 2026. New interagency model risk guidance replaced SR 11-7, and it leaves generative AI outside its scope.

What bankers use AI for

Bank staff use AI in two different ways. Some use general AI assistants to draft and summarize. Others work with AI models that make or support decisions, such as fraud scores. The second kind falls under model risk management. The first kind mostly falls under the bank's own AI use policy.

• Credit memos: drafting the first version of a memo from the financial spreads and borrower documents, for the credit officer to check and sign.
• Compliance: summarizing a new rule or exam finding, and drafting a policy update or a procedure change.
• Alert review: writing a first-pass summary of a fraud or AML alert for an analyst to confirm or close.
• Customer service: drafting replies to routine questions for a staff member to review.
• Operations: extracting fields from loan documents and checking them against the system of record.

In each case a person signs the output and is responsible for it. The credit officer who signs a memo answers for its figures, whatever tool drafted them.

What changed in model risk guidance in 2026

On April 17, 2026, the Federal Reserve, the OCC and the FDIC issued revised guidance on model risk management. The Federal Reserve issued it as SR 26-2. It supersedes SR 11-7, the 2011 guidance that banks had built their model risk programs on. It also supersedes SR 21-8 on BSA/AML models (Federal Reserve). The Federal Reserve says the guidance is most relevant to banking organizations with over $30 billion in total assets.

The guidance takes a risk-based approach. It states that it does not set enforceable standards (Federal Reserve). It also excludes generative and agentic AI.

Generative AI and agentic AI models are novel and rapidly evolving. As such, they are not within the scope of this guidance.

The same footnote says a bank's own risk management and governance practices should guide controls for tools the guidance does not cover. In practice, a bank employee's use of a generative AI assistant is governed by the bank's internal policy. Training on that policy matters more than any outside credential.

The AI credentials bankers can take

Each entry comes from the issuer's own page. Prices are as listed by each issuer in October 2026.

• ABA AI in Banking Workshop (American Bankers Association): $595 for ABA members, $795 for non-members. It runs for four weeks, with self-paced modules and a live session each Thursday. Topics are AI types in banking, prompting and agents, guardrails for safe use, and everyday uses such as fraud detection. The page does not describe an exam or a certificate (ABA).
• AI in Banking Applied: Risk & Compliance (American Bankers Association): $775 to $795 for members, $995 for non-members; four weeks, blended. It is aimed at risk, compliance, audit, legal and governance staff. It covers AI risk, governance and controls, third-party AI due diligence, and exam readiness (ABA).
• Risk and AI (RAI) Certificate (GARP): for the 2027 exam windows, $550 to $650 for individual members and $650 to $750 for non-members, depending on when you register (GARP). The exam has 80 multiple-choice questions over four hours, offered in April and October. GARP estimates 100 to 130 hours of study (GARP).
• Artificial Intelligence Fundamentals Certificate (ISACA): US$120 for members, US$144 for non-members; a two-hour, remotely proctored multiple-choice exam with a 65% pass mark. It suits bank IT, security and audit staff who need a baseline (ISACA).

What each credential proves

• The two ABA courses prove that you completed a structured four-week program built for banks. Neither page describes a graded assessment.
• The GARP RAI Certificate proves that you passed a long proctored exam on AI risk, model governance and responsible AI. It carries the most weight for risk and model validation roles.
• The ISACA certificate proves basic knowledge of AI concepts and risks, tested by exam.

None of these tests whether you can draft a sound credit memo with AI and catch its errors. That skill is the one a credit or compliance manager will see every day.

How to choose

1. Read your bank's AI use policy first. It tells you which tools you may use and on what data.
2. Frontline, operations and lending staff: the ABA AI in Banking Workshop covers banking uses and guardrails in four weeks.
3. Risk, compliance and internal audit staff: the ABA Risk & Compliance course covers governance and third-party AI review.
4. Model risk, validation and quantitative staff: the GARP RAI Certificate is the exam-based option.
5. IT and security staff: the ISACA certificate gives a tested baseline for US$120 to US$144.

What an employer can verify

For the ABA courses, ask the candidate for the completion record. Then read the course page to see what the course required. For an exam credential such as RAI or the ISACA certificate, ask for the issuer's record of the pass and its date.

To see the skill itself, give a short task in the interview. Hand the candidate a one-page borrower summary with two planted errors. Ask them to draft a memo paragraph with AI and say what they checked. A team AI use policy and the guide on keeping AI use safe set out what a bank should expect.

Where Emeri fits

Emeri sells an AI proficiency certification, so weigh this comparison with that in mind. Anyone can take the free check, a 15-minute check of how you work with AI. Emeri certification is assessed on real work tasks and reported as a band from E1 to E4 (bands). It is not specific to banking, and it does not cover model risk management.

For related roles, see AI certifications for risk managers and AI certifications for compliance.