A leaked AI API key: someone else runs the attack, you get the bill

An API key for an AI service, if leaked from a script or the website's chatbot, can be used to fund attacks against other companies, and the bill ends up with the key's owner.

Bitsight researchers monitored 16 criminal forums and found a tool, MONST3R-AI, marketed as an “autonomous security testing platform.” The tool does not have its own AI model: the buyer pastes in stolen keys from OpenAI, Anthropic, or Google, sets a budget, and sends an AI agent to scan targets. The usage is charged to the account from which the key was stolen.

For a small company, these kinds of keys come up more often than you might think: the chatbot on the website, an automation tool that summarizes emails or fills out documents, or a script written by an external contractor. Often, no one in the company knows exactly where the key is or who still has it.

An API key is like a password with your company's credentials attached. Whoever has it can make requests on your behalf without logging into any account or going through two-factor authentication.

What you need to do:

  • Find out where your company keeps its AI API keys and who has access to them, including those created by the agency or freelancer who built your website or set up your automations.
  • Set a monthly spending limit and a usage alert on the provider’s account; a sudden spike in your bill is the first sign.
  • Use a separate key for each application, and never store them in emails, shared documents, or published code. If one gets leaked, revoke it and issue a new one.

Sources: Help Net Security | https://www.helpnetsecurity.com/2026/10/16/stolen-ai-api-keys-autonomous-scans/

This article was generated with AI assistance.

Request a quote

← All news