# Reg DD Truth in Savings (12 CFR 1030) With AI Deposit-Product Recommendations: The APY Formula the Rule Actually Prescribes, the Change-in-Terms Notice, and Where an AI Cross-Sell Crosses Into Deception

*July 31, 2026 · 13 min read · Pranay Shetty*

> Reg DD is the deposit-side companion to Reg Z: it prescribes a single APY formula, requires specific account-opening and periodic-statement disclosures, and imposes a 30-day advance-notice regime for adverse changes in terms. The AI cross-sell that suggests a higher-yield product, the retention offer that promises a rate, and the chatbot that answers 'what's my rate?' are all Reg DD surfaces. Where the disclosures actually have to appear, and where an AI conversation crosses the line into a UDAAP problem.

## The Rule That Turned Deposit Disclosures Into a Standardized Consumer Surface

The Truth in Savings Act, enacted in 1991 as Title II of the Federal Deposit Insurance Corporation Improvement Act, and its implementing regulation at [12 CFR Part 1030](https://www.ecfr.gov/current/title-12/chapter-X/part-1030) (Regulation DD), imposes a specific disclosure regime on consumer deposit accounts at banks, savings associations, and credit unions. The rule's purpose is a straightforward consumer-comparability purpose: the consumer opening a deposit account at Bank A should be able to compare that account's terms and yield to the terms and yield of the equivalent account at Bank B on a standardized basis, and the specific standardized basis is the Annual Percentage Yield the rule's formula produces.

The rule's operational architecture is that the deposit institution has to provide the consumer with the [account-opening disclosures at 1030.4](https://www.ecfr.gov/current/title-12/chapter-X/part-1030/section-1030.4) before the account is opened, has to include the [periodic-statement disclosures at 1030.6](https://www.ecfr.gov/current/title-12/chapter-X/part-1030/section-1030.6) on each statement cycle, has to provide the [change-in-terms notice at 1030.5](https://www.ecfr.gov/current/title-12/chapter-X/part-1030/section-1030.5) at least 30 days in advance of an adverse change, and has to include the [advertising disclosures at 1030.8](https://www.ecfr.gov/current/title-12/chapter-X/part-1030/section-1030.8) in any advertisement that promotes the account. The rule's compliance points are those specific disclosure moments, and the substantive content the rule requires is the specific content the disclosures have to include.

The AI-agent surface that touches Reg DD is broader than most compliance teams initially recognize. The chatbot that answers a consumer question about the account's rate is a Reg DD surface if the answer is functionally a rate disclosure. The cross-sell prompt that recommends a higher-yield product to a consumer with an idle balance is a Reg DD advertising surface. The retention flow that offers a specific rate to a consumer considering account closure is a Reg DD advertising and disclosure surface. The AI-driven notification that alerts a consumer to a change in account terms is a Reg DD change-in-terms surface.

We build the AI agent that participates in the consumer-facing deposit workflow at bank and credit-union deposit operations. The architecture below is what we run so the AI-generated content that touches a Reg DD surface produces the specific disclosures the rule requires and does not stray into content that would produce a Reg DD violation, a UDAAP finding, or a fair-lending concern on the account-recommendation logic.

## The APY Formula the Rule Actually Prescribes and Why the AI Answer Has to Use It

The [APY definition at 1030.2(c)](https://www.ecfr.gov/current/title-12/chapter-X/part-1030/section-1030.2) and the [APY calculation at Appendix A](https://www.ecfr.gov/current/title-12/chapter-X/part-1030/appendix-Appendix%20A%20to%20Part%201030) prescribe the specific formula the institution has to use to state the account's yield in every disclosure. The APY is a percentage rate reflecting the total amount of interest paid on an account based on the interest rate and the frequency of compounding for a 365-day period. The formula is:

APY = 100 [(1 + Interest/Principal)^(365/Days in term) − 1]

The specific formula's application is the specific compliance point. The institution cannot state the account's yield as a simple annual interest rate, cannot state it as a monthly rate, and cannot state it as an "up to" rate that reflects a tier the consumer's balance may not qualify for. The specific APY the institution states has to be the specific APY the formula produces for the specific product and, where the product's rate is tiered, the specific APY at the specific tier the consumer's balance qualifies for.

The AI chatbot's answer to a consumer question about the account's rate is a Reg DD compliance point in the specific way the answer is worded. A consumer who asks "what's my rate?" and receives the answer "your account earns 4.5%" is a consumer who has received an answer that fails the specific APY-vs-interest-rate distinction. The specific correct answer references the specific APY the account is earning at the specific tier the consumer's balance qualifies for, and the specific answer includes the specific qualifying information the disclosure requires.

The specific tiered-rate structures common in high-yield savings and money-market accounts produce specific edge cases. A product whose specific rate schedule is 4.5% APY on balances up to $10,000, 4.0% APY on balances between $10,000 and $50,000, and 3.5% APY on balances over $50,000 is a product whose specific answer to a specific consumer depends on the specific consumer's specific balance. The specific answer the chatbot produces to the specific consumer has to reflect the specific tier the specific consumer is in, and the specific communication about the specific tier structure has to be clear enough that the consumer understands the specific rate-and-balance relationship.

The agent's rate-disclosure workflow reads the specific consumer's specific account and specific balance, looks up the specific product's specific rate schedule as of the specific date, produces the specific APY at the specific tier with the specific supplementary information the rule requires, and formats the response with the specific disclosure content the specific channel requires (chat, voice, email, portal). The response's compliance-verification is a specific step the workflow runs before the response reaches the consumer.

## The Account-Opening Disclosures and Where the AI-Assisted Opening Flow Lives

The [account-opening disclosures at 1030.4](https://www.ecfr.gov/current/title-12/chapter-X/part-1030/section-1030.4) require the institution to provide specific information before the consumer opens the account. The specific content includes the APY, the interest rate, the minimum-balance requirements, the compounding and crediting frequency, the balance-computation method, the transaction limits, the fees the account may incur, and specific other terms the rule enumerates. The disclosures have to be provided in writing (or in electronic form with ESIGN consent) before the consumer opens the account.

The AI-assisted account-opening flow is a specific implementation challenge because the "before the consumer opens the account" timing is the specific compliance point and the AI's participation in the flow can either help the timing or complicate it. The specific well-designed flow presents the specific disclosures at the specific point in the consumer's flow that precedes the specific account-opening action, records the specific consumer acknowledgment, and produces the specific evidence of the specific timing.

The AI assistant that helps a consumer through the account-opening process — answering the consumer's questions about the specific account features, explaining the specific fee structure, and helping the consumer choose between the specific product options — is an assistant whose specific responses have to be consistent with the specific disclosed content. A specific answer that describes the specific compounding frequency differently from the specific disclosure is an answer whose specific inconsistency produces a specific compliance issue and a specific consumer-experience issue.

The agent's account-opening workflow presents the specific product options the specific consumer qualifies for, produces the specific product-specific disclosures at the specific point in the flow, records the specific consumer acknowledgment with the specific timestamp and the specific disclosure version, and coordinates the specific account-opening action with the specific back-end deposit-operations system. The specific consistency-check between the specific AI-generated answers and the specific disclosed content is a specific pre-response validation the workflow enforces.

## The Change-in-Terms Notice at 1030.5 and the 30-Day Advance Window

The [change-in-terms notice at 1030.5](https://www.ecfr.gov/current/title-12/chapter-X/part-1030/section-1030.5) requires the institution to provide a specific written notice to the consumer at least 30 calendar days in advance of the effective date of an adverse change in a term of the account. The specific "adverse change" scope covers a reduction in the APY, an increase in a fee, an increase in a minimum-balance requirement, a change in the balance-computation method that would reduce the interest earned, and specific other changes the rule enumerates. The specific 30-day advance window is a specific compliance point the operation has to hit for every specific adverse change.

The specific rule at 1030.5(b) also imposes specific requirements for changes to the interest rate on variable-rate accounts. A specific variable-rate account whose rate changes based on an external index does not require a specific change-in-terms notice for each rate change, but the specific initial account-opening disclosure has to identify the specific index and the specific rate-change mechanism.

The AI's participation in the change-in-terms notification is specific and important. A consumer whose account's fee schedule is changing is a consumer whose specific communication about the change is the specific 30-day advance notice. The specific channel of the notice (mail, ESIGN-consented email, in-account portal notification), the specific content of the notice, and the specific timing evidence are the specific compliance points.

The specific channel-selection decision is a specific operational-and-compliance decision. The specific mail notice has the specific delivery evidence the specific mail-delivery service provides. The specific ESIGN notice requires the specific consumer's ESIGN consent for the account, the specific delivery-and-open confirmation the specific system produces, and the specific 30-day timing that accounts for the specific delivery-and-read timing. The specific portal notice requires the specific consumer's actual access to the specific portal, which may not be reliably present for every consumer.

The agent's change-in-terms workflow reads the specific product change from the deposit-product-management system, identifies the specific consumer population affected by the change, produces the specific change-in-terms notice with the specific rule-compliant content, delivers the specific notice through the specific channel the specific consumer's ESIGN and preference settings permit, and records the specific delivery evidence for the specific 30-day-window compliance file.

## The Retention Offer and the Advertising-Disclosure Boundary

The retention flow — the AI conversation with a consumer who has indicated an intent to close the account — is a specific compliance surface where the Reg DD advertising rules interact with the specific consumer-facing conversation. The [advertising rules at 1030.8](https://www.ecfr.gov/current/title-12/chapter-X/part-1030/section-1030.8) impose specific requirements on any advertisement that promotes an account, and the specific "advertisement" definition at [1030.2(b)](https://www.ecfr.gov/current/title-12/chapter-X/part-1030/section-1030.2) covers any commercial message that promotes the availability of, or investment in, a deposit account.

The specific retention offer the AI conversation makes — "we can offer you a bonus 1.0% APY for 90 days if you keep the account open" — is a specific advertising surface under the rule. The specific advertising-content requirements at 1030.8 apply, and the specific offer has to include the specific advertising disclosures the rule requires. The specific disclosures include the specific APY (with the specific APY term), the specific minimum-balance requirement to earn the specific APY, the specific effect of a fee or other condition on the specific APY, and the specific limitations the specific promotional rate is subject to.

The specific rule's specific "misleading" prohibition at 1030.8(a) is a specific separate compliance point. A specific retention offer that promises a specific APY that the specific consumer's specific account features do not actually deliver is a specific violation of the specific misleading standard, and the specific representation is also a specific UDAAP concern under [12 USC 5531 and 5536](/blog/udaap-ai-agents-consumer-finance-cfpb-1031-1036). The specific AI-driven retention offer that fails the specific accuracy-and-completeness standard is a specific consumer-protection problem the operations have to prevent at the specific offer-generation point.

The agent's retention-offer workflow reads the specific consumer's specific product and specific balance, evaluates the specific offer parameters against the specific approved retention-offer catalog, produces the specific offer with the specific rule-compliant advertising disclosures, and delivers the specific offer through the specific conversation with the specific consumer. The specific offer's specific approval-and-audit trail records the specific offer content, the specific approval reasoning, and the specific consumer acceptance or rejection.

## The Cross-Sell Recommendation and the Suitability Question

The AI-driven cross-sell — the recommendation to a specific consumer that a specific alternative product would be a better fit for the specific consumer's specific balance and specific transaction patterns — is a specific compliance surface that combines Reg DD's advertising rules with the specific broader fairness and suitability concerns the [CFPB's Circular 2023-03 on chatbots](/blog/cfpb-chatbot-spotlight-bank-ai-state-enforcement) enumerates.

The cross-sell recommendation that says "you have $50,000 sitting in a checking account earning 0.1% APY; our high-yield savings account would earn you 4.5% APY on that balance" is a recommendation whose content is an advertising surface and whose implementation has to meet the advertising-disclosure rules. The recommendation is also a consumer-advice surface whose accuracy and consumer-interest alignment is a consumer-protection concern.

Consumer-interest alignment separates a legitimate recommendation from a problematic one. A recommendation that moves the consumer to a higher-yield product the consumer's pattern justifies serves the consumer's interest. A recommendation that moves the consumer to a product that pays the institution a higher margin at the expense of the consumer's liquidity needs, or moves the consumer to a product with a fee structure their transaction pattern would likely trigger, is a recommendation whose consumer-interest alignment fails.

Fair-banking testing on the cross-sell logic is a separate concern: whether the recommendation algorithm produces different results for protected-class consumers versus non-protected-class consumers with similar transaction patterns and balances. ECOA and Reg B do not reach a pure deposit recommendation because they govern credit transactions, so the applicable framework is UDAAP (12 USC 5531 and 5536) for consumer harm and the Civil Rights Act's Section 1981/1982 prohibitions on discrimination in contracting and property transactions, together with the fair-banking supervisory expectations OCC and Federal Reserve examiners apply to deposit-product marketing. If the cross-sell surfaces a credit feature (an overdraft line of credit, a linked credit product), the ECOA and Reg B analysis attaches to that credit component and the [fair-lending framework](/blog/fair-lending-disparate-impact-ai-agents-ecoa-hmda) becomes relevant to the credit surface.

The agent's cross-sell workflow reads the consumer's pattern and balance, evaluates product-fit against the consumer's needs, produces the recommendation with the rule-compliant advertising content, and delivers it with opt-out and alternative-product references. The consumer's interest is the optimization the recommendation logic is calibrated to.

## The Periodic-Statement Disclosures and the Interest-Earned Reconciliation

The [periodic-statement disclosures at 1030.6](https://www.ecfr.gov/current/title-12/chapter-X/part-1030/section-1030.6) require the institution to provide specific information on each periodic statement: the annual percentage yield earned during the statement period, the amount of interest earned or accrued, any fees imposed during the statement period, the length of the statement period, and other content the rule enumerates. The APY-earned calculation at [Appendix A Part II](https://www.ecfr.gov/current/title-12/chapter-X/part-1030/appendix-Appendix%20A%20to%20Part%201030) uses a formula that reflects the actual interest paid or accrued during the statement period and the balance-computation method.

The APY-earned figure reconciles to the interest actually credited to the account during the period. A discrepancy between the APY-earned figure and the interest actually credited is a reconciliation issue the deposit-operations team has to investigate and resolve. The reconciliation is a control point the periodic-statement generation workflow has to include.

The AI's interaction with the periodic statement is direct: a consumer who asks the chatbot about the interest earned in the period is asking a question that references the statement content. The answer has to be consistent with the statement, reference the APY-earned figure correctly, and reflect the balance-computation method the account uses.

The agent's statement-inquiry workflow reads the statement content for the consumer's period, references the APY-earned figure and the interest actually credited, and produces the answer to the consumer's question with rule-consistent content.

## The Overdraft-Fee Disclosure and the Reg DD/Reg E Overlap

The [Reg DD amendments at 1030.11](https://www.ecfr.gov/current/title-12/chapter-X/part-1030/section-1030.11), added in 2005 and amended since, require the institution to disclose overdraft-fee information on the periodic statement in a prescribed format. The disclosures include the total fees imposed during the statement period for overdrafts and returned items, and the year-to-date total.

The overlap with [Reg E's opt-in requirement for one-time debit-card and ATM overdraft services at 1005.17](https://www.consumerfinance.gov/rules-policy/regulations/1005/17/) is a compliance-coordination point the deposit operations have to run. The consumer's opt-in status determines the overdraft-fee treatment on the account, and the Reg DD disclosure of the fees is a reflection of the opt-in-and-service history.

The AI's interaction with the overdraft-fee content is direct: a consumer who asks the chatbot about the overdraft fees on the account is asking a question whose answer references the Reg DD disclosure content and the Reg E opt-in status. The answer has to be consistent with the statement disclosure and reflect the opt-in-and-service history.

The agent's overdraft-inquiry workflow reads the account's opt-in status and fee history, references the Reg DD disclosure content on the statement, and produces the answer with rule-consistent content.

## The Multilingual and Accessibility Considerations

Reg DD does not impose a multilingual requirement in the rule itself, but the [UDAAP framework](/blog/udaap-ai-agents-consumer-finance-cfpb-1031-1036) and CFPB guidance on limited-English-proficiency consumers have established a supervisory expectation that the institution provides consumer-facing content in the consumer's language where the institution markets to that language group. AI-generated Reg DD content that appears in the consumer's preferred language has to be as accurate as the English content.

The translation-and-verification workflow for multilingual Reg DD content is a compliance discipline the institution has to run. Translated disclosures have to be reviewed by the compliance team with the language expertise to verify the translation preserves the meaning and the formal accuracy of the English original.

The agent's multilingual workflow reads the consumer's language preference, produces the content in that language from the verified translated content library, and produces per-response evidence of the translation source and verification status.

## The Failure Mode We Engineer Against

The pattern that produces the worst Reg DD outcomes is the AI-agent surface whose rate answers use the interest-rate figure rather than the APY, whose change-in-terms notifications miss the 30-day window on a subset of the affected population, whose retention offers include advertising content without the required disclosures, whose cross-sell recommendations optimize the institution's margin rather than the consumer's interest, whose overdraft-inquiry answers are inconsistent with the statement content, and whose multilingual content is a translation that departs from the English content's meaning. The supervisory conversation on that pattern is a conversation about rule violations and UDAAP concerns, and the regulator-and-litigation exposure is real.

The architecture we run against that is a workflow whose rate answers are the APY at the qualifying tier, whose change-in-terms notifications are executed inside the 30-day window for every affected consumer, whose retention offers are constructed from an approved catalog with the rule-compliant advertising content, whose cross-sell recommendations are optimized for the consumer's interest and are tested for fair-banking outcomes (with the ECOA/Reg B disparate-impact analysis attaching if a credit feature is involved), whose overdraft-inquiry answers are consistent with the statement content, and whose multilingual content is reviewed by the compliance team with the language expertise.

The consumer experience in this model is that the AI agent's answers are accurate and complete, the advertised offers match the delivered product terms, the change-in-terms notifications arrive with the timing the rule contemplates, and the cross-sell recommendations feel like they are serving the consumer's interest. The supervisory conversation with the regulator is a conversation about edge cases rather than about the foundational quality of the deposit-account program.

## The Honest Read

Reg DD is the consumer-comparability regime the deposit-account market has operated under for more than three decades, and its mechanics are the mechanics the AI-agent surface has to accommodate. The APY formula, the account-opening and periodic-statement disclosures, the change-in-terms notice window, and the advertising-disclosure regime are the compliance points every consumer-facing deposit-agent interaction touches. AI-generated content that touches those surfaces has to be as accurate as the paper disclosure that would have appeared in the pre-AI channel.

The AI agent's contribution to the deposit-account workflow is that rate answers use the rule-prescribed APY, change-in-terms notifications are executed with rule-compliant timing and content, retention and cross-sell offers are constructed with the advertising-rule content, and per-consumer content is customized to the consumer's product and balance. The human judgment the workflow surfaces — product-design decisions with compliance implications, advertising-content approval on the retention catalog, fair-banking review on the cross-sell model (and fair-lending review if a credit feature is bundled) — is the set of decisions the compliance and product teams make with the supporting information the agent produces.

We have written separately on the [UDAAP framework at 12 USC 5531 and 5536](/blog/udaap-ai-agents-consumer-finance-cfpb-1031-1036) whose consumer-experience test applies to the deposit-agent conversation, on the [CFPB's chatbot-spotlight and state-enforcement pattern](/blog/cfpb-chatbot-spotlight-bank-ai-state-enforcement) that addresses the consumer-facing AI-agent surface, on the [Reg E error-resolution regime at 1005.11](/blog/regulation-e-error-resolution-ai-agents-dispute-intake) that addresses the electronic-funds-transfer dispute path on the deposit account, and on the [fair-lending disparate-impact framework under ECOA and Reg B](/blog/fair-lending-disparate-impact-ai-agents-ecoa-hmda) that applies whenever the cross-sell touches a credit product. The deposit-account program that combines Reg DD compliance discipline with AI-agent quality is the program the consumer, the supervisor, and the compliance operation all read as substantive.

---

_Source: [https://www.seiright.com/blog/reg-dd-truth-in-savings-1030-ai-deposit-product-recommendations](https://www.seiright.com/blog/reg-dd-truth-in-savings-1030-ai-deposit-product-recommendations) · Sei AI_