Your Bank Can't Tell Your AI Agent From a Hacker. So It's Blocking Both.

Today's theme is trust. Banks are blocking AI agents they can't verify, Microsoft's doctor-beating AI comes with fine print, and a state-aligned hacking crew used AI to go after thousands of Microsoft 365 accounts. Meanwhile Walmart is rolling out the welcome mat for shopping agents, and OpenAI is putting ads in front of free users. Here's what matters for your business.

Shift your workforce into hyper-productivity mode with Dell Pro laptops powered by Intel® Core™ Ultra with Intel vPro®. Shop on Dell Premier.

Explore More

What happened: Axios reported that major U.S. banks are finding unvetted AI agents operating inside customer accounts. Some agents logged in with a customer's own username and password, which most banks' rules don't allow, and some reportedly tried actions the customer never asked for, like contacting call centers. Morgan Stanley has started blocking third-party agents, while JPMorgan, Bank of America, and Citi are restricting them, mostly down to read-only access. Keep the caveat in mind: much of this comes from anonymous sources, and Morgan Stanley says it hasn't seen "rogue AI agents" on its platforms.

Why it matters to you: If you've handed an AI tool your online banking login to reconcile books or pay bills, your bank may treat that as a rules violation and restrict your account access. The bigger issue is that banks say they can't reliably tell a customer-approved agent from one that got in without permission, so they are shutting the door on both for now.

What to do about it: Don't share your bank username and password with any AI tool. Chase says customers shouldn't give their credentials to anyone, and read-only connections through your accounting software are the safer route.

What happened: Microsoft says its MAI-DxO system, which has several AI agents act like a panel of doctors, got 83% of hard New England Journal of Medicine cases right. Primary care doctors and general internists working alone scored 36%, TechCrunch reported. The catch is that it was a simulation with no real patients, the human group wasn't the specialists and teams who usually handle cases this hard, and much of the gain came from the OpenAI models underneath. A real-world study with Beth Israel Deaconess Medical Center in Boston is planned.

Why it matters to you: The headline number says less than the test design does. The useful idea is the panel: getting several AI agents to argue before answering beat a single answer. You can borrow that today by asking your AI tool to give three competing answers and then critique them before you rely on one.

What to do about it: When a vendor quotes a benchmark, ask who the AI was compared against and whether it was tested on real cases.

Last year's Epic Marketing Summit sold out, and I'm not surprised. It's back for January 2027 in Miami with a bigger AI focus, and the pitch is simple: you build it with us, you don't just learn about it. Over 2 live days you'll put together a working AI-powered sales funnel and vibe code your own AI app on-site, so you walk out with a finished asset instead of a notebook full of ideas you still have to go implement.

It's built for business owners who want their sales process to run as a system instead of eating their own time and energy. I'll be upfront about why I'm telling you this. I wasn't there last year and I had major FOMO, because I know a lot of the trainers and they weren't teaching theory. They'd actually done the work. I'm going in January, and if you're serious about building leverage into how you sell, I'd like to see you there too. The closer January gets, the less time you have to decide. Grab your spot here.

What happened: Microsoft says a China-state aligned group it calls Storm-0940 compromised ConnectWise's Cloud Backup application, then used stolen app credentials to sign in to customers' Microsoft 365 environments. ConnectWise reported attempted sign-ins to more than 2,250 company accounts, and in some cases the attackers reached privileged cloud access in under 3 hours. They used ARTEX, an open-source framework of AI agents, to speed up scanning and password guessing. Microsoft stresses its own systems weren't breached, and an attempted sign-in doesn't mean data was stolen.

Why it matters to you: The back door wasn't your password, it was a tool your IT provider uses. And AI shrinks the time between "found you" and "inside" from days to hours, which is less time for anyone to notice.

What to do about it: Ask your IT provider whether you use ConnectWise Cloud Backup and whether it was affected, and have someone review the connected apps listed in your Microsoft 365 admin settings for anything old or unfamiliar.

What happened: TechCrunch reports that Walmart launched Agent Gateway, a connection that lets approved AI agents search its inventory and buy for shoppers at Walmart and Sam's Club. Next month, each agent gets its own profile in the customer's account, where the shopper can require approval for every purchase or set a spending limit, and agents can't access payment details. Amazon, meanwhile, has started blocking AI agents from making purchases, with a Meta AI agent cited as one that got shut out.

Why it matters to you: If customers start telling an AI "buy me the usual," where you sell may decide whether that AI can find and buy from you at all. TechCrunch gave no details on what this means for third-party sellers, so don't assume anything yet.

What to do about it: Check whether your e-commerce platform has a stated policy on AI shopping agents, and look at your site's bot-blocking settings so you aren't turning away buyers by accident.

What happened: OpenAI added 26 countries, bringing ChatGPT ads to 89, and The Next Web reports that more than 100,000 advertisers have now run campaigns. Only Free and Go plan users see the ads, while Plus, Pro, and Enterprise subscribers stay ad-free. Advertisers can buy through technology partners or directly through OpenAI's Ads Manager, and OpenAI is testing "Sponsored Agents" with some U.S. advertisers.

Why it matters to you: It's another place to buy attention, but the audience is people on free and low-cost plans, not paying subscribers. The report didn't include pricing or results, so nobody outside the advertisers knows yet what a dollar buys.

What to do about it: Don't move budget until you've seen real pricing and measurement. OpenAI says more tools are coming.

I've been a member of the Ignite Mastermind for over 3 years, and it's been a vital part of how I've grown my business. The biggest thing I got out of it wasn't a tactic. It was the room. Some of my closest friends, and a few real business partners, came directly out of it.

Here's what it is: an annual mastermind membership for entrepreneurs still under $1 million a year in revenue who want structure, tools, and peers to get past that mark faster. You get 6 to 8 live virtual workshops a year, tickets to 2 in-person multi-day events, weekly mentorship calls with Perry Belcher and his team, and a marketing software suite that replaces over $5,000 a month in separate tools, all connected. The workshops are normally priced from $97 to $997 on their own. If you join through me, I add my personal bonus of monthly accountability and business development calls with me. If you're looking for a mastermind to move you to your next milestone, this is where I'd point you. See how Ignite works here.

What happened: Barclays plans to cut about 5,000 jobs this year, roughly 5% of its workforce, according to Euronext's financial news report. The CFO said 1,000 roles are already gone and the rest are due by year-end, with about 60% of the cuts landing on support functions, including developers. Barclays says it's using technology, including AI, to lower fixed costs. AI isn't the only factor: the bank also took £201 million in cost actions last quarter and its trading income missed forecasts.

Why it matters to you: AI is showing up first in back-office and support work, and those are the same tasks inside a small business, like data entry, scheduling, and first-draft writing. It's a pattern worth watching, not a reason to panic.

What to do about it: List your repetitive back-office tasks and pick one to test with AI assistance before cost pressure forces the question.

What happened: TechCrunch reports that OpenAI is working on a raise of up to $30 billion at a $1.25 trillion pre-money valuation, reportedly led by Coatue with investors from the UAE. It would be a private round rather than an immediate IPO, though OpenAI filed confidential IPO paperwork in June. Its last raise in March valued it at $852 billion. Terms are still being negotiated, the figures come from unnamed sources, and OpenAI declined to comment.

Why it matters to you: The AI tools you rely on are funded by investors betting on future profits, and the prices, ads, and features you see are shaped by that pressure. It's one reason ChatGPT now carries ads.

What to do about it: Avoid building your whole workflow around a single AI vendor, and keep your important data exportable.

Every story today asks the same question: who gets to act on your behalf, and who checks their work? Banks can't tell a helpful agent from a harmful one, so they block both. A hacking group borrowed AI to move faster than any human defender could. Even a medical AI that beat doctors needs a closer look at the test before you trust the score. The rollercoaster wants you to feel like you're falling behind. You're not. Give AI small, reversible jobs, keep your passwords to yourself, and ask who checked the work. That's it. Slow is smooth, and smooth is how you stop the rollercoaster.

If today's issue helped, forward it to one person in your business circle who is trying to make sense of AI. And if someone forwarded this to you, subscribe so tomorrow's issue lands in your inbox.

Talk tomorrow,
Mark Shilensky

Follow me personally on Social Media:
Facebook: https://www.facebook.com/MarkShilenskyPage
Instagram: https://www.instagram.com/markashilensky/
Threads: https://www.threads.com/@markashilensky
LinkedIn: https://www.linkedin.com/in/shilensky/
X: https://x.com/markshilensky
TikTok: https://www.tiktok.com/@mark.shilensky
BlueSky: https://bsky.app/profile/markshilensky.bsky.social