Key Takeaways
- ClearScore launches ACBP, the world's first open protocol for AI agents to conduct regulated credit broking
- The protocol defines a "responsibility separation model" where agents handle dialogue while brokers retain regulatory control
- Gartner predicts 40% of financial institutions will deploy AI agents by end of 2026, making industry standards urgent
A "New HTTP for Finance" Launches in London
World's first Agentic Credit Broking Protocol sets a standard for financial services bots.
www.machine.newsOn April 7, 2026, UK fintech ClearScore unveiled the Agentic Credit Broking Protocol (ACBP) in London. It is the world's first open protocol enabling AI agents to conduct credit broking while remaining compliant with financial regulations. The specification has been published on GitHub, targeting industry-wide adoption.
Founded in 2015, ClearScore operates across the UK, Australia, New Zealand, Canada, and South Africa, serving over 26 million users with more than 200 financial institution partners. CEO Justin Basini positioned ACBP as fulfilling the same role for credit broking in the AI agent era that web browsers played during the internet era.
Why a Protocol Is Needed Now
In traditional financial services, regulated firms controlled every user touchpoint. Whether an in-person advisor or a website, the "first interaction surface" always belonged to the broker.
The rise of AI assistants is dismantling this assumption. Consider a user asking ChatGPT or Claude, "Should I take out a car loan?" That conversation might begin in a chat app, pass through a comparison tool, and be handed off to an embedded assistant. The regulated firm no longer controls the first interaction surface.
Without a protocol, these journeys are opaque — to the broker, to the lender, and to any regulator who later needs to understand what occurred.
Gartner predicts that 40% of financial services firms will be using AI agents by end of 2026. The UK's FCA is also investigating the long-term impact of AI on retail financial services through the Mills Review, led by Executive Director Sheldon Mills. The absence of a protocol is no longer a theoretical risk — it is an urgent practical concern.
ACBP's Core Design: Separating "Interaction" from "Responsibility"
At the heart of ACBP lies a simple but innovative concept: enabling "interaction and responsibility to travel separately."
In practice, an AI agent collects a user's income, expenditure, and existing debt through natural dialogue, then hands this over to a broker as structured data. The broker evaluates the user's circumstances, selects appropriate products, and obtains offers from lenders. The AI agent then conveys those offers to the user and guides them through the non-regulated parts of the application.
Throughout this entire process, the AI agent never needs to become a regulated entity. The broker retains regulatory control and an evidence trail is preserved. This shares a philosophical kinship with the Trusted Agent Protocol and Agent Pay frameworks that Visa and Mastercard are developing in the payments space, but ACBP is notable for venturing into credit broking — a more heavily regulated domain.
However, challenges surfaced almost immediately after publication. A developer filed 11 issues on GitHub, raising concerns about prompt injection attacks and model drift — the degradation of a model's predictive accuracy over time. Most critically, the claim that there is "no structured mechanism for the broker to signal that a user has been identified as vulnerable" points to a fundamental question about FCA compliance.
Cape Town Development Hub and Global Expansion
Alongside the ACBP launch, ClearScore announced the establishment of an AI-driven credit innovation hub in Cape Town, South Africa. The company already serves over 6.5 million users in South Africa — one in eight citizens — and plans to strategically expand its local team of approximately 24 people.
Development of what Basini calls the "universal API programme" will be based in Cape Town. ACBP is designed to work across jurisdictions, with the explicit intent of leapfrogging the interoperability challenges that have plagued open banking.
Implications for E-Commerce Operators
While ACBP's most direct impact falls on financial institutions and credit brokers, e-commerce operators cannot afford to ignore this development.
Merchants that integrate BNPL or deferred payment options into their checkout flows may eventually see ACBP-compliant AI agents brokering credit products at the point of sale. If a consumer tells their AI assistant, "I want to buy this product on the best possible terms," the agent could retrieve optimal credit offers via ACBP and seamlessly complete the transaction.
Meanwhile, US regulators continue to apply existing financial rules to AI systems rather than creating new frameworks, leaving open questions about how autonomous agents fit into regimes designed for human decision-makers. E-commerce operators should verify how their payment and credit partners are responding to agentic commerce standardization efforts and begin preparing accordingly.
Looking Ahead
ClearScore's ACBP signals that the agentic commerce wave has reached beyond payments into the more heavily regulated territory of credit broking. The fact that vulnerability concerns surfaced on GitHub immediately after publication arguably validates the open protocol approach.
The key focal points going forward are the recommendations from the FCA's Mills Review, expected this summer, and how broadly ACBP gains adoption among financial institutions. If Gartner's prediction holds and 40% of financial firms deploy AI agents by late 2026, the race to establish the "common language" for these agents is already underway.
