As AI Takes Control of Finances, a New Blueprint for Trust Emerges

AUSTIN, Texas – June 15, 2026 – For years, we’ve grown accustomed to artificial intelligence working in the background of our financial lives, offering recommendations, flagging potential fraud, or powering chatbots. But a fundamental shift is underway. AI is moving from advisor to actor—a change that carries both immense promise and unprecedented risk. Recognizing this pivotal moment, the non-profit Responsible AI Institute (RAI Institute) today announced the launch of TrustX for Finance, a major initiative aimed at creating a safety framework for AI systems that can independently execute financial transactions.

Backed by founding members including U.S. Bank and NatWest Group, the program is a direct response to the rise of “agentic AI”—systems that don't just generate outputs but can plan, decide, and take action with delegated authority. As these autonomous agents begin to initiate payments and execute workflows, the financial industry is grappling with a stark reality: the governance models built for the last generation of AI are no longer sufficient. TrustX for Finance aims to build the new foundation before the technology outpaces our ability to control it.

The Dawn of the Autonomous Banker

The transition from advisory to agentic AI is not a distant sci-fi concept; it’s happening now. A recent report found that 62% of financial services firms have already deployed AI agents, and a staggering 93% of those have granted them some level of autonomy. These systems are moving beyond customer service and cybersecurity to operate at the very heart of finance.

This new class of AI can, for example, manage corporate treasury functions, execute trades based on market analysis, or autonomously approve and issue loans. The potential for efficiency is enormous, but so is the “blast radius” if something goes wrong. As the RAI Institute’s press release notes, this risk is amplified by two key trends. First, many firms are deploying AI through third-party vendors and SaaS platforms, often with limited visibility into the agent’s behavior or reach. Second, advanced AI models are increasingly capable of interacting with a company’s internal tools and data, creating pathways for misuse, compromise, or misalignment on a massive scale.

“Financial institutions cannot approve autonomous AI for production using governance models built for static systems,” said Manoj Saxena, Founder and Executive Chairman of the Responsible AI Institute. “As AI begins to initiate payments, execute workflows, and act with delegated authority, the industry needs a shared way to classify risk, enforce boundaries, and prove systems are operating as approved.”

This isn't just about preventing rogue trades. The risks are deeply embedded in the data these systems learn from. Algorithmic bias can lead to discriminatory outcomes in lending and credit scoring, while the complexity of “black box” models makes it difficult for firms to explain decisions to regulators or customers—a significant hurdle in a tightly regulated industry.

A Regulatory Race Against Time

While the industry forges ahead, global regulators are racing to keep pace. The European Union’s landmark AI Act, which classifies AI used for credit scoring as “high-risk,” will see its key provisions for financial services take effect in August 2026. In the United States, the SEC has proposed rules targeting conflicts of interest in AI-driven investment advice. Just this month, the Financial Stability Board (FSB) released its own long-awaited governance framework, explicitly warning that agentic AI could “undertake illegal, unethical, or unauthorized actions without human approval or oversight.”

Despite these efforts, a significant gap remains. A 2026 report found that 43% of financial regulators are not even collecting data on AI adoption, highlighting a fundamental challenge in creating informed policy. The industry is moving faster than the oversight bodies can effectively monitor, creating a vacuum that initiatives like TrustX for Finance are designed to fill.

“As consumers and businesses begin using AI systems that can act on their behalf, financial institutions need a common assurance framework,” noted Dr. Samuel Assefa, a Senior Vice President at U.S. Bank, in the announcement. “Preparing for new trends and the inevitable expansion of Agentic AI use cases is critical.”

Building a Verifiable Blueprint for Trust

TrustX for Finance is not another theoretical white paper; it’s a hands-on program designed to provide a structured and verifiable path for deploying autonomous AI. The framework is built on a core principle: governance must follow what AI systems are allowed to do, not just how they are built.

The initiative provides a comprehensive toolkit for institutions, structured around three domains: Build (for internally developed systems), Buy (for assessing third-party AI), and Protect (for securing enterprise systems exposed to AI agents). At its center is a “proving ground,” a controlled sandbox environment where banks can test and validate autonomous AI systems against shared criteria before they are deployed in the real world.

Within this environment, institutions can classify systems into risk tiers based on their autonomy, decision authority, and potential reach. This allows them to apply controls that are proportional to the risk and aligned with regulatory expectations. Crucially, the process is designed to generate audit-ready evidence, providing a defensible record for internal compliance teams, external auditors, and regulators.

“TrustX for Finance comes at a critical moment for our industry,” said Dr. Paul Dongha, Head of Responsible AI & AI Strategy at NatWest Group. “As financial services organizations begin deploying agentic AI, we must move quickly but responsibly — assessing the risks of this powerful new technology, embedding robust controls before deployment, and proving those controls hold in production.”

To foster industry-wide adoption, the outputs of the working group will be made available through the TrustX Open AI Registry. This openly licensed resource will provide a shared schema for risk classification, agent blueprints, and control policies, allowing the entire sector to build upon a common foundation.

This model is not entirely new for the RAI Institute. It builds on the success of TrustX Health, a program launched in the UK in December 2025 to create a similar assurance pathway for agentic AI in healthcare. That experience in another highly regulated, high-stakes industry provides a credible precedent. Initial workstreams for the finance initiative will focus on autonomous commerce and payments, collaborating on real-world use cases to test and refine the shared assurance criteria.