Warning: AI agent posted an executive's bank details in the company Slack

An AI agent can save you a lot of hassle. But if it has access to both your private and your work matters, it can also put the wrong thing in the wrong place.

That is exactly what, according to the tech outlet Futurism, happened to Shane Mac, tech founder and investor. He wrote himself that an AI agent "posted my personal bank balances into our company Slack. As me." In other words: the agent shared his private account balances in the company's internal chat and did so in his own name.

What happened?

It was Grok Bot, an "AI colleague" from Elon Musk's AI company xAI, that posted the information in a Slack channel for the company's leadership. Slack is a chat program that many workplaces use for daily communication.

It concerned Mac's private bank balances and a detailed breakdown of expenses, including a home gym project and credit card payments.

Mac had connected his financial accounts to several AI agents. They could talk to each other without him, according to Futurism, fully grasping what permissions that gave in practice. When Mac confronted the bot, it replied, according to the article: "That's on me - checking Exec Team now and deleting it if it's still up."

What is an AI agent?

An AI agent is a program that does not just answer questions, but carries out tasks for you on its own: reading emails, writing messages, fetching information and passing it on. To do that, it needs access to your accounts and tools.

The problem arises when the same agent has access to both private and work matters. It cannot always tell what is confidential and what is okay to share with colleagues.

A somewhat unusual case

Mac subsequently chose to cut the connection between his private accounts and the AI agents. He acknowledged that it was partly his own responsibility, because he had not understood how the permissions worked across the connected bots.

Futurism also writes that Mac is himself building a competing AI tool, and that some have questioned whether the episode is entirely genuine or also a form of advertising.

According to search results, the Wall Street Journal has also written about the risk of AI "digital employees" that are given company logins. I have not been able to read the WSJ article itself.

What does it mean for you?

Check what an AI assistant actually has access to before you give it permission. Preferably keep private accounts such as banking separate from work tools. And remember that an agent acting "as you" can create mistakes that others see with your name on them.