You can’t blink these days without seeing another article or discussion about artificial intelligence. There’s certainly no shortage of places to learn about AI, and we’re not looking to constantly add to all that noise.
But in our role as Personal CFOs to our clients, there is one development we’ve been paying attention to that we thought was worth sharing, especially as we’re sure you’ll start seeing more headlines about it.
It’s the emergence of personal AI agents that can do things on your behalf, from reading and responding to emails to shopping and making reservations.
What has caught our attention isn’t so much what these tools can do. It’s the level of access to your personal information and accounts that some of these tasks may require.
For example, I recently received a marketing email from ChatGPT inviting me to:
“Connect your finances and let ChatGPT review where your money goes and build a budget for what’s next.”
I can certainly understand why that might sound appealing. And I suspect some of you have received similar invitations or are already experimenting with these tools.
But it also raises some questions worth thinking about before connecting your financial accounts.
A little about AI agents
Some of you may already be familiar with tools like Muse and Instinct. Others may be hearing about them for the first time.
First, a caveat. We’re financial planners, not technology or cybersecurity experts. My understanding of these tools comes from reading about them and talking with people who know a lot more about the technology than I do. I’m sure some of you reading this have considerably more expertise in this area.
But here’s how I understand the basic distinction.
Many of us first encountered ChatGPT and similar tools as something we could ask questions. We typed in a question, and AI gave us an answer.
Personal AI agents take that a step further. Rather than just telling you how to do something, they can actually go do it for you.
That might mean managing appointments, responding to emails, booking travel, or even making purchases for you – not just suggesting what you should do, but actually taking steps on your behalf.
To do some of those things, though, they need access to the websites, applications, and bank accounts or credit cards you use.
And when those accounts involve giving them direct access to your finances, that gives me pause.
I’m personally not comfortable sharing my banking passwords or providing broad access to my financial accounts to these tools.
That doesn’t mean there won’t be valuable uses for them. I expect there will be. But I think there’s a meaningful difference between asking AI to help you understand your spending and giving it your username and password for full access to your bank accountin order to figure that out for you.
Why we’re bringing this up now
This isn’t just a hypothetical concern.
On October 22, 2026, Axios reported that major U.S. banks were detecting third-party AI agents accessing customer accounts. In some instances, the agents appeared to be attempting actions beyond what customers had instructed them to do.
Reuters also covered the story, and several financial institutions have reportedly begun restricting how third-party AI agents can access their systems.
It’s important to distinguish between an agent accessing an account in a way the bank hasn’t approved and an agent actually stealing money or making an unauthorized transfer. The reporting doesn’t establish that customers lost money in these incidents.
Still, I think it’s worth paying attention when banks themselves are raising concerns about how these tools are accessing accounts and what they might be able to do once they’re inside.
A few things we’d think about before experimenting
- Be very careful about sharing passwords.
I personally would not give my online banking username and password to a personal AI agent.
Some applications use secure connections that allow you to authorize limited access without handing over your password. That’s different from giving a tool your login credentials or allowing it to operate through your existing banking session.
Either way, I’d want to understand what information it can access and what it’s allowed to do.
And I’d be particularly cautious about anything that can initiate transactions or move money.
- If you’re curious, consider starting somewhere that doesn’t involve your primary financial accounts.
If you’re considering AI-assisted shopping, one option may be a virtual payment card with a limited spending amount. That can help limit financial exposure, although it doesn’t eliminate privacy or other risks.
The general idea is to avoid giving a new technology access to more than it needs.
- Understand what you’re authorizing—and whether you can take that permission back.
Before connecting any service, I’d want to know whether it can simply view information or whether it can also make changes, send messages, or initiate transactions.
Can you approve actions before they happen? Can you see what the agent has done? Can you easily disconnect it?
These are reasonable questions to ask of any third-party financial application, not just AI agents.
And if you don’t understand the answers, I’d be inclined to hold off.
A few other security reminders while we’re on the subject
Even if you have no interest in experimenting with AI agents, this is a good opportunity to revisit some basic practices that can help protect your financial information.
- Use a password manager. A reputable password manager can generate and store strong, unique passwords so you don’t have to remember dozens of them.
- Let the password manager generate your passwords. Your cat’s name, even with a few numbers and symbols added, probably isn’t the best choice. Randomly generated passwords are much harder to guess, and there’s no reason to reuse the same password across different accounts.
- Use passkeys when they’re available. You may be seeing more websites offer passkeys as an alternative to passwords. They generally allow you to sign in using your device’s fingerprint, facial recognition, or PIN and provide stronger protection against phishing. Passkeys can be securely managed through reputable dedicated password managers or supported device and platform managers.
- Turn on multifactor authentication. Particularly for your financial accounts, email, and password manager. It provides another layer of protection if someone obtains your password.
- Pay attention to account alerts and connected applications. Transaction notifications can help you identify unexpected activity. It’s also worth occasionally reviewing which outside applications have access to your accounts and removing connections you no longer use.
My perspective
I don’t feel any urgency to be among the first to connect my personal financial accounts to these newer tools.
I suspect we’ll see continued improvements in how AI applications access information, what permissions we can give them, and how financial institutions work with them.
In the meantime, I think a little patience and a healthy amount of curiosity are a reasonable combination.
For those who want to learn a little more, here are a few resources that caught our attention:
