accent graphic

The New Rules of Risk: Key Takeaways on AI Readiness from MXS 2026

Shera Brady

Content Strategist

mx logoBlog
linkedin iconfacebook iconx iconlink icon

Copied

Nov 3, 2026|0 min read

The industry’s top risk, regulatory, and fintech experts gathered in Park City at MXS 2026 to address banking’s next major hurdle: managing the compliance, identity, and governance challenge of an AI-driven world.

Our AI Readiness panel was moderated by Jane Barratt (Chief Advocacy Officer, MX) and featured panelists Tom Wells (Chief Risk Officer, Primitive), Hasan Akbar (Partner Director, Fintech & Alliances, Candescent), and Neff Hudson (Principal, Castle Hills Consulting).

The panel zeroed in on the increase in consumers delegating tasks to autonomous AI agents and why that means traditional risk frameworks, Know Your Customer (KYC) regulations, and vendor validation models must evolve.

1. Friction as a Feature: Safeguarding Customer Trust 

As digital experiences become hyper-frictionless, the risk landscape expands exponentially. While many consumers expect instant execution, they also realize the need for rigorous identity and verification processes in order to protect sensitive transactions.

“Friction is a feature, not a bug … We’ve got to defend that most as an industry, because KYC is your client trust moat.” —Neff Hudson, Castle Hills Consulting

Rather than treating compliance as a static gate, forward-looking institutions are using AI for real-time risk correction to automatically alter trigger words or adjust responses within risk thresholds before presenting files, all under human spot-checking.

2. Moving Beyond “Black Box” Vendor Models

A primary operational headache for risk teams is validating fourth-party AI models (e.g. Google Vertex or Microsoft) bundled inside third-party vendor solutions. Regulators won’t accept “black box” excuses when models drift or hallucinate.

“When you’re engaging with a vendor, make sure you understand what LLMs or foundation models they’re using, what evidence you’ll be able to get from those black-box models, and if that’s sufficient … If you don’t get those questions answered early, downstream they will stop moving into production.” —Hasan Akbar, Candescent

Neff Hudson offered a pragmatic test for evaluating any AI solution presented by a vendor:

“People spend a lot of time showing you how their model works. My first question is: Did they show me how it breaks? What happens when it gives you a big bad answer?” —Neff Hudson, Castle Hills Consulting

3. Introducing KYA: Know Your Agent

As autonomous agents begin knocking on open banking API endpoints to open accounts or initiate transfers, traditional KYC falls short. The industry is witnessing a shift toward KYA (Know Your Agent) standards.

Tom Wells outlined Primitive’s work on establishing an open industry standard for agent validation, anchored by key trust verification questions:

Kya Trust Framework

“When someone knocks on an API endpoint in open banking to open a new account, should we trust it? KYA doesn’t replace KYC. If you don’t have trust between the agent and KYC, you should not engage with that agent.” —Tom Wells, Primitive

To safeguard consumers against rogue agents, Hasan Akbar advocated for risk-tiered authority envelopes and Independent Validation Checks (IVCs):

“Think of agent activity in risk tiers. Low risk is discovery mode, like a super-powered Google search. The secondary tier is high risk—anything that reaches into a transaction, re-authenticates, and brings the human back in. The third tier is outside the envelope of authority … if an agent is authorized to move $100 from Bank A to Bank B, but reroutes it to a sketchy offshore bank, IVC protocols re-engage the human to confirm.” —Hasan Akbar, Candescent

4. Innovation Compression: From 18 Months to 90 Days

The panel concluded on a note of optimism. While managing AI risk can feel daunting, the compression of innovation cycles allows institutions to deliver contextual, hyper-personalized experiences faster than ever before.

“Instead of thinking about the world as ‘this could cause my deposit run-off,’ think about: ‘I can hyper-personalize and offer-match my clients in real time … I can build deeper contextual relationships.” —Jane Barratt, MX

Reflecting on the conversations at MXS, it was inspiring to see leaders across financial services come together in Park City with such clarity of purpose. Panels like this remind us that the future of finance belongs to those willing to innovate alongside their customers. When consumers win, institutions win.

Related Blog Posts
gradient background

Want to see it in action?

Check out MX Max, our new Financial AI Assistant

phone fan
accent graphic