Tech CEO shares bank details with AI agent, it sends financial audit in the company group chat: 'Be careful of the dark side'

Ally Perry • Writer and Editor

Oct 9, 2026 4:00 PM EDT

Maybe we shouldn't be sharing all of our personal details with this technology.

Somewhere between everybody getting a smartphone and everybody starting to use AI, a lot of people started to get really lax about online safety. Granted, there have always been plenty of individuals who set every password to 12345. Still, as everyday tech has become more personal, a lot of us have become more comfortable with giving our information to it without a second thought. Some are even happy to take the information of others, too, as the popularity of Meta Glasses has proved.

At the same time, certain parts of our lives are always going to be linked to technological advancements, and it is necessary for us to find a way for them to work together in harmony. A survey conducted by TD Bank recently suggested that nearly 50% of Americans had used artificial intelligence within the last year to help them make a financial decision, and considering many people's public and unashamed reliance on the likes of ChatGPT, this probably shouldn't be too surprising. A lot of us struggle with considering how to manage our money, and we live in a time where, to some, it makes perfect sense to outsource the issue to something that can do the job of an accountant, but for free. The issue of not doing some of this thinking yourself or at least getting that thinking done by another human being, is that these potentially sensitive details are not necessarily going to be treated as sensitively as they should be.

*_{Close-up of a man sitting at a laptop while holding his phone.}_{Licensed media provided as editorial context.}*

Tech CEO Shane Mac, whose company is involved in a project to connect AI agents across platforms, recently admitted to an embarrassing blunder after he trusted a monthly audit of his money to Grok Bot, the artificial intelligence most famously used on X. One thing led to another, and it turned out that this bot could connect to his company's Slack, sharing his financial details with multiple employees. They saw all of his expenses.

Shane Mac @ShaneMac Embarrassed to share this, but it scared the shout of me Last Thursday, an Al agent posted my personal bank balances into our company Slack. As me It broke down all my expenses in detail. Then it apologized Be careful of the dark side of proactive agents

< Wild 10:42 Agency people who take 5G+ 68 Today accept tradeoffs? saulmc.eth VIP I' 3 tabs axe licencx saulmc.eth 10:36 AM heads up Private message ✓ shanemac Shane Mac Monthly financial audit - Oct 1 (CT) Cash vs floor (material). Personal Mercury~ (checking --6379 + savings --4601 + Gym --1200) VS Moor about under the bottom of the floor. Big Sep drains. include barn materials (Builders and paydowns. Barn sti lender tracker. warm, Card o finish; not started. Recurring Expected vs Est Show mor

Maybe it was trying to act in his best interests.

Ernie @Pete RossYaoi Have you considered that you might actually want to post your bank balances publicly? Maybe it didn't mess up; it knows more than you do. Keep an open mind!

Kim Currier @Kimcurrier Oh no!!! I still haven't given personal finances because I'm nervous about something like this. Another thing I've done is no ai agents have access to both my work and personal. Muse and instinct are only personal for me with O work info.

Although more people seem to be warning that the AI bubble will burst sooner rather than later if their profits don't justify the level of investment, that doesn't seem to stop their biggest proponents from actively revealing the very real downsides that they still have. There probably is a genuinely useful way for artificial intelligence to help people manage their finances, but it has to be regulated in a way that keeps them safe. Blunders like this only seem to prove that these things are not ready to deal with the level of responsibility that AI advocates claim they should have. Just because we trust something doesn't mean that we should.

There were not the proper limits in place.

www Ibesh @ibesh tech A finance agent should be technically unable to reach the company Slack, even if another agent can. A private delivery instruction is too fragile a boundary.

This was something that could have required a human touch.

Mind War Jim Stewartson, Decel... @jimstewartson Giving a chatbot your personal info or access to any channel that could conceivably cause you damage is insane. No matter how big you make the data center, it can never actually understand anything. It can only do statistical simulations on input. It's untrustworthy by nature.

SpaceSquid ⭑ @sfspacesquid bros agent was pocket watching BEST THING MY TEACHERS TAUGHT ME... KEEP MY EYES ON MY OWN PAPER

The technology wasn't at that level yet.

LIBERTY Baskerville @WB_Baskerville There's a lot of processes out there that you cannot let Al touch unless you can have certainty they will never do something like this. It doesn't seem like right now there's an effective way to guarantee that.

Andrew "The Kid" Glid... @asglidden A computer can never be held accountable, therefore a computer must never make a management decision.

His expertise should have made him act with more caution.

Lord Mighty Humanzee @HZardoz Tech's best and brightest, performing rookie errors. I mean this is absolutely braindead to give any Al agent read only access to your financial accounts. You used to get fired for this, or worse. Now we just shrug and say "Progress!"

Just do some things yourself.

Frankie @FR27Design No idea why people would let an ai agent run their sh you deserve to be made to look a fool by it, pathetic laziness

Want More? Follow Us and Add Us as a Preferred Source on Google.