October 28, 2026

Can a Stranger Use Your AI Chatbot on Your Dime? 5 Checks to Run This Week

In October, two incidents let others spend other people's AI credits. Five plain checks to make sure yours isn't open to anyone.

Sources last checked: October 28, 2026, 11:15 AM ET.

Short answer

If your business uses an AI chatbot or assistant, check who can reach it and what it can spend. In October, reports said strangers chatted with another company's AI agents on its customers' credits, and unauthorized charges followed at a coding tool where users had stored AI keys. Five checks help: list your AI tools, test who can reach them, set spending limits, limit access, and keep a person approving money and deletions.

If you've added an AI chatbot or assistant to your business this year, it may be doing more than you think. Many of these tools answer customers, book jobs or read email, and every message they handle costs something. In October, two real incidents showed what happens when someone else can reach that. In one, the agents were simply findable. In the other, stored AI keys were exposed. Either way, someone else ran up the bill.

What happened

A search feature made AI agents findable. Red Team Partners, a security-testing firm, reported on October 26 that a search function in German software firm Tobit's chayns platform "made SideKick AI agents findable for a short time," and "strangers started chats with them on customers' credits." Tobit says about 100 of roughly 10,000 installations were targeted, no personal data left, and the credits were refunded. In one customer's account, the IT team "had not switched on any AI features" but still found credits spent by someone they couldn't trace.

Exposed AI keys turned into surprise charges. TechCrunch reported on October 19 that attackers broke into an older system at OpenCode, an AI coding tool, affecting 1.4% of users of its online console, where some users stored the keys they use to pay for AI services. Some users reported unauthorized charges. TechCrunch says some reported "charges in the hundreds of dollars" and "a few described totals above $1,000." OpenCode told affected users to revoke and replace their keys, and said refunds were underway.

The lesson from both: if someone else can reach your AI tool or your AI keys, they can run up your bill.

5 checks you can do this week

You don't need to be technical for these. You need some time and access to the admin pages of the AI tools you pay for.

  1. Make a list of every AI tool you use, including the ones that came free. Write down each chatbot, assistant and AI feature, and what it connects to: email, calendar, customer list, payments. Include AI features your software vendors switched on inside products you already use. The Tobit customer above had not switched on any AI features and still found credits spent.
  1. Check who can reach it. Open your website chat or assistant link in a private browser window while logged out. If it answers, the public can use it too. Make sure that's intended. Internal assistants shouldn't show up in search results, directories or shared links.
  1. Set a spending cap and an alert. In each AI account's billing page, set a monthly limit if the tool offers one, and an email alert well below it. Red Team Partners puts it simply: set a hard limit "so a night of bot traffic stops at a number you chose." Without one, nothing stops the charges until you notice them.
  1. Give it only the access its job needs. Verizon's October 21 small-business guide says to "give an AI agent only the information and permissions needed for its task." Its example is that an appointment-booking agent may need your calendar and customer contacts, but not payroll, bank accounts or permission to issue refunds. Remove any connection the tool doesn't need.
  1. Keep a person in charge of money and deletions. Verizon also advises that AI agents "should require human approval for actions outside the routine you tested." Turn on approval steps for refunds, payments, deleting records and new user accounts. While you're in the admin settings, remove any user or account you can't explain.

One more thing: if you store AI account keys anywhere online, know where they are. If a service that holds them reports a breach, revoke the keys and replace them with each AI provider, as OpenCode told its users to do.

What High Power can do

We can sit down with you for a one-time look at the AI tools your business uses: who can reach them, what they're connected to, and whether spending limits and approval steps are in place. You make the changes in your own accounts, and we walk you through them. For this review, we never ask for your passwords or log into your AI accounts.

Not sure where you stand? Call or text (757) 869-4742, or email contact@highpower.tech.

This post was drafted with AI and approved for publishing by a person at High Power Technical Solutions.

Sources

  • Red Team Partners (a security-testing firm), "Strangers Chatted With a Software Firm's AI Agents on Its Customers' Credits. Its CEO Says Attacks Have Never Been This Cheap." (Oct 26, 2026): https://redteampartner.com/field-notes/tobit-sidekick-ai-agents-exposed/
  • TechCrunch (Aditya Mehta), "Hackers hit AI coding tool OpenCode, running up charges on users' AI accounts" (Oct 19, 2026): https://techcrunch.com/2026/10/19/hackers-hit-ai-coding-tool-opencode-running-up-charges-on-users-ai-accounts/
  • Verizon (Tia Hopkins), "Already using an AI agent? Make these 3 checks before it does more" (Oct 21, 2026; the author was compensated by Verizon): https://www.verizon.com/about/news/ai-agent-security
High Power Technical Solutions · Hampton Roads, Virginia