Welcome to AI for FIs, from Dixon Strategic Labs. Each week, this newsletter curates critical developments in agentic AI and explains why they matter for your credit union.

AI agent traffic grew 7,851% in 2025. Bots now account for 52% of all internet activity. A year ago, one in every 200 website visits came from an AI agent. Now it's one in 31.

Those agents are shopping. And some financial institutions are already visible to them. Salesforce measured $262 billion in U.S. holiday sales influenced by AI agents. When agents picked financing at checkout, they surfaced lenders with structured, machine-readable product data. Credit unions that had rates and terms in that format showed up. The ones with loan rates buried in PDFs didn't.

Solaris, a licensed European bank, fired 20% of its staff and rebuilt around agents. Citizens Financial is investing $300 million over three years in AI-assisted service, including routing half its call center volume through agents. Visa launched Agentic Ready in Europe, where 21 issuers (Barclays, Revolut, Santander, and others) are running live agent-initiated transactions in controlled environments.

1, 2, 3, let’s dig in.

Search Engine Land · Mar 26, 2026

A HUMAN Security report found automated web traffic grew 23.5% in 2025, compared to 3.1% for human traffic. AI agent traffic grew nearly 8,000% year over year. 77% of observed agent activity hit product and search pages, and some agents reached checkout flows.

Why it matters: AI agents are comparing products and completing transactions on behalf of consumers. They choose FIs whose product data they can read. Credit unions that make rates, terms, and eligibility machine-readable now will be discoverable in a channel that didn't exist 18 months ago.

FinTech Magazine Article · Mar 23, 2026

Salesforce reported AI agents influenced $262 billion in U.S. holiday sales in 2025, or 20% of all retail sales. When agents surfaced financing options at checkout, lenders with structured, accessible product data were the ones that appeared.

Why it matters: Credit unions already compete on rates. Those rates also need to be visible to the agents doing the comparison. Structured data and APIs are the minimum. Agent commerce is early enough that showing up is still a differentiator, and the CUs that move first have a real window.

FinTech Futures · Mar 26, 2026

German BaaS fintech Solaris is cutting 80 roles (about 20% of staff) as part of a pivot to AI-native banking. The company plans to deploy AI agents for operational processes while humans handle governance. The restructuring follows a €140M funding round led by Japan's SBI Group.

Why it matters: Solaris sits between tech company and licensed bank. Their reduction in force will put pressure on other licensed banking providers to explain why they aren't restructuring around agents. For what it's worth, I believe in re-org over delete-org as a response to this transition. The past issue’s feature — “Your Credit Union’s Next AI Move Should Be Boring” — touches on this.

A documentary just hit theaters from the producers of Everything Everywhere All at Once and Oscar-winning director Daniel Roher. Roher, a father-to-be, sits down with Sam Altman, Dario Amodei, Demis Hassabis, and a collection of AI researchers and asks each of them: “is now a terrible time to have a kid?”

None of them can promise him it isn't. The film maps the economic, workforce, and security implications of AI from the ground up, then zooms into the racing dynamics between labs and between countries. Tristan Harris, Ilya Sutskever, and Yuval Noah Harari are among folks that also sit for interviews.

8.2 on IMDb. 89% on Rotten Tomatoes. I saw it Friday. Check it out. Take your board!

Anthropic's Claude Mythos Leak

Anthropic, the company behind Claude, confirmed it is testing a new AI model called Claude Mythos after an internal configuration error left nearly 3,000 company documents exposed on the public internet. Among them: a draft announcement describing the model as the most powerful Anthropic has ever built. The company called it "a step change" in capability.

According to Axios, Anthropic has been privately warning senior government officials that Mythos makes large-scale cyberattacks much more likely in 2026. The leaked documents described the model as far ahead of any other AI system at finding and exploiting software vulnerabilities.

The credit union angle is direct. In November 2025, Anthropic disclosed that a Chinese state-sponsored group had used an earlier, less capable version of Claude to autonomously attack roughly 30 organizations, including financial institutions. The AI handled 80-90% of the operation on its own. Mythos is a generation beyond the model used in that attack.

Separately, Anthropic also accidentally published the source code of its popular coding tool, Claude Code, just days later. Two major leaks in one week from the same company. Cybersecurity stocks fell sharply on the news. A Dark Reading poll found 48% of cybersecurity professionals now rank AI agents as the top attack vector for 2026.