Federal Reserve official: Generative AI could advance financial inclusion — with strict guardrails

Michael, an agency official at the Federal Reserve, told a public discussion that generative artificial intelligence could help increase financial inclusion for low- and moderate-income people if it is deployed with intentional policy and operational safeguards. He cited examples such as AI‑assisted English-language support and task-focused tools that can make workers more productive and potentially boost wages.

"If properly deployed, artificial intelligence can help low income people in lots of different ways," Michael said, highlighting uses from language translation to targeted budgeting support. He warned, however, that the technology poses distinct risks in financial services — including threats to privacy, the potential to entrench bias in credit decisions, and risks to market integrity if AI accelerates herd behavior in trading.

Moderator Tim, who has worked on financial-capability initiatives, asked how to build trust among people who are currently less likely to use frontier generative AI. Michael said AI can both help and harm trust: it can provide tailored, nonjudgmental responses that increase appropriate trust in institutions, yet the same tools can enable fraud or cyberattacks if misused. "AI might, help overcome some of those trust barriers," he said, while adding the caveat that the technology can also be used by bad actors.

Both speakers discussed the evolution of financial education toward "financial capability" — brief, decision‑focused help delivered at the moment it matters — and Michael said generative AI could accelerate that model by delivering individualized advice at lower cost. He pointed to research and practice showing that users are more likely to rely on technology they find useful, delivered by a trusted provider, and to adopt tools that leave them feeling competent rather than judged.

Michael also framed AI's potential to increase household "slack" — the small buffers of time, money or attention people need to avoid a cascade after common shocks such as car repairs or lost income. Referencing his prior work on scarcity, he said AI could help households anticipate problems, plan, and smooth cash flow through savings, insurance or credit where appropriate.

On systemic concerns, Michael called for guardrails in several areas: stronger privacy protections for data used by AI systems; safeguards against automated bias in credit and other decisions; mechanisms to ensure value alignment so AI systems act consistently with individuals' and societal values; and oversight to prevent AI-driven trading from amplifying runs or facilitating algorithmic collusion. "So these are some of the examples," he said, urging that policy and governance keep pace with new AI capabilities.

Kate Hyung of the Boston Fed asked which improvements in financial inclusion might be undermined by AI. Michael answered that financial services have made important progress over the past 30 years to design products for low- and moderate‑income households, and that AI could help accelerate that progress by improving product design and measurement — but only if institutions put appropriate controls in place to avoid amplifying existing inequities.

The session concluded with Michael's central takeaway: generative AI can help reduce inequality and improve financial inclusion, but only if stakeholders act with "intentionality and purpose." Moderator Tim closed by thanking Governor Barr for joining the conversation.