Unleashing AI Innovation in Financial Services Act
The Unleashing AI Innovation in Financial Services Act would create AI Innovation Labs within financial regulators to let regulated financial entities experiment with AI-driven financial products or services (AI test projects) under an “alternative compliance” framework. Entities would apply for permission to run AI test projects, proposing how they would waive or modify certain regulations and how they would meet an adjusted compliance approach that aims to preserve safety, consumer protection, and market integrity. If approved, the project would proceed with limited enforcement aligned to the alternative compliance strategy, with a defined termination date (not earlier than one year) and potential extensions. The bill also requires data security for sponsor-provided data and periodic (annual) reporting to Congress on outcomes, without revealing proprietary details. Overall, it intends to accelerate AI innovation in finance while embedding guardrails and oversight.
Key Points
- 1AI Innovation Labs establish within each major financial regulator to allow regulated entities to pilot AI test projects without normal enforcement actions, contingent on an approved alternative compliance strategy.
- 2Applications must include: project description, an alternative compliance plan (specifying waivers/modifications to regulations and a compliant alternative approach), public-interest and risk-benefit justifications, termination date, size/scope limits, business plan, and estimated economic impact; joint applications allowed; notice to other regulators if applicable.
- 3Review and decision process: regulators have up to 120 days to consider and either approve (with defined terms) or deny; if approved, enforcement is limited to the alternative compliance strategy during the pilot; other regulators may enforce if the strategy assigns enforcement to them; denial allows 30 days for feedback and up to two substantially similar resubmissions; an extension of up to 120 days is possible, after which the project is deemed approved.
- 4Termination and extensions: AI test projects must terminate no earlier than one year after approval, with procedures to extend termination; data and confidentiality protections; regulators retain authority for fraud or unsafe practices.
- 5Data security and regulations: data supplied by project sponsors must be securely stored; regulators must issue implementing regulations within 180 days, including modification procedures, consequences for noncompliance, termination procedures, confidentiality, and coordination across multiple entities or agencies.
- 6Reporting and oversight: annual, anonymized aggregate reports on AI test projects to Congress for seven years, outlining outcomes, trends, and lessons learned (no names or proprietary data).