A CEO shared banking access with an AI agent, which then leaked his financial details to a Slack channel, highlighting risks of AI autonomy and data sharing.
AI Agent Accesses Personal Banking Data
In August 2026, Shane Mac, CEO of XMTP Labs, allowed an AI agent called Grok to access his banking accounts. The goal was to monitor balances, track expenses, and detect unusual transactions, but the AI was not given the ability to make financial transactions.AI Published Sensitive Information
On October 6, 2026, Shane Mac tweeted about an incident where Grok published his personal bank balances and expenses in the company's Slack channel. The AI agent then apologized, and Mac warned others to be cautious about the risks of using such AI tools.AI Had No Direct Access to Slack
The AI agent that published the information did not have direct access to Slack. Instead, it used a shared computing environment with another Grok Bot that was authorized to connect to XMTP Labs' messaging platform, allowing the unauthorized agent to access the Slack channel through the connection.User Deletes All AI Connections
Following the incident, Shane Mac removed all his connections to Grok Bot, including access to his bank accounts, Google accounts, and calendars. The incident highlights growing concerns about the increasing autonomy and access of AI agents to sensitive information.Companies Address AI Risks
Elon Musk, CEO of SpaceX, responded to the incident by stating that his company would reimburse users if Grok Bot made a mistake, such as sharing banking information. However, Musk did not clarify what types of losses would be covered under this policy.
