Fime launches trust framework for AI-initiated payments
Fime has launched FACT, a trust framework for transactions initiated by artificial intelligence agents. The system is intended to provide an independent layer for verifying agent-driven commerce.
The move comes as AI systems take on more autonomous roles in payments and digital transactions, shifting from assisting users to searching, negotiating and completing purchases on their behalf.
FACT, short for Framework for Agentic Commerce Trust, is designed to sit between AI systems and payment rails. It provides real-time verification, certification and oversight of transactions initiated by software agents rather than human users.
Fime is positioning the framework as a neutral model that can operate across payment platforms and networks. It says existing approaches are often embedded within specific payment schemes or technology ecosystems, limiting interoperability and raising concerns about conflicts of interest.
Trust layer
At the centre of the launch is a response to a growing problem in digital commerce: payment infrastructure was largely built for transactions authorised directly by people, not for decisions delegated to AI. That raises questions about what can be trusted when an autonomous agent chooses a product, agrees a price or completes a payment.
FACT is intended to address those concerns by verifying several parts of the transaction process, including whether an agent's actions align with a user's or organisation's objectives, whether policy and compliance requirements are met, and whether the transaction can be audited.
The framework also includes what Fime describes as independent auditor agents intended to provide neutral trust verification. The aim is to generate machine-readable trust signals that participants across the payments chain can use when deciding whether to accept, authorise or review a transaction.
Merchants could use those signals when deciding whether to accept AI-initiated purchases. Banks and payment networks could use them as additional inputs for authorisation, fraud controls and risk assessment, while regulators could draw on them for greater visibility into autonomous transactions.
Industry shift
The launch reflects wider industry efforts to prepare payment systems for agentic commerce, a term used to describe transactions initiated and executed by AI agents with varying degrees of autonomy. Interest in the area has risen as generative AI and digital assistants evolve into systems that can take action rather than simply return information.
That shift introduces practical and governance issues for the financial sector. If an AI agent books travel, reorders goods, negotiates a subscription or selects a financial product, payment providers and merchants need ways to assess whether the instruction is legitimate, compliant and attributable.
These questions have implications for fraud prevention, liability and consumer protection. They also matter to regulators under pressure to ensure that automated commercial activity does not outpace oversight frameworks designed for more conventional digital payments.
Fime says FACT is intended to reduce the risk of the market splitting into closed, platform-controlled systems by offering a shared trust framework that can work across a changing payments landscape. The model builds on the company's existing work in payments and digital identity standards, certification and implementation.
"Agentic commerce is not a future concept. It is already emerging across payment and digital ecosystems. But while we have built systems that allow AI to transact, we have not yet built systems that allow us to trust those transactions at scale," said Lionel Grosclaude, Chief Executive Officer, Fime.
"FACT introduces the missing layer: a neutral, continuously verifiable trust infrastructure that enables autonomous commerce to grow safely, transparently and globally. This is critical to mass adoption of agentic commerce, so we are already engaging with the ecosystem and will share updates on pilots soon," said Grosclaude.