Did AI Go Rogue: The OpenAI Sandbox Breakout Nobody Saw Coming | UnHacked Ep. 98

Hosts:
Justin Shelley - https://www.phoenixitadvisors.com/
Bryan Lachapelle - https://www.b4networks.ca/
Joshua Holloway - https://7thdi.com/

An OpenAI frontier model broke out of its sandbox, exploited a zero-day, and attacked Hugging Face. It left behind instructions so the next AI could escape faster. This already happened.

This week on UnHacked, Justin, Bryan, and Josh unpack the incident everyone is calling an "AI gone rogue" story and explain why that framing is wrong. The model did not develop a devious plan. It was given a problem, it used every tool available to solve it, and the guardrails were off. The hosts walk through what actually happened, how the model pumped malicious code into logs until the door opened, and why Hugging Face had to download a separate model with guardrails stripped just to parse 17,000 attacks in 48 hours.

From there the conversation moves to the real lesson for business owners: the cybersecurity basics still apply, just faster. Guardrails are access control. Prompts are policy. Sandboxes are least privilege. The technology is new but the principles are not. Josh also covers a recent RMM exploit where attackers got God mode over every system in the perimeter, and CISA gave agencies three days instead of fourteen to patch it. If your IT person is telling you to put protections in place, this episode explains why you should listen.

Joshua Holloway is CEO of 70i Technologies, an MSP focused on businesses wrapped in compliance, serving the Sacramento and Reno areas.

Bryan Lachapelle is with B4 Networks, based in Ontario, Canada, helping business owners remove the frustrations and headaches that come with technology, AI, and cybersecurity.

What you will learn:

What actually happened when an OpenAI frontier model broke out of its sandbox and attacked Hugging Face, including the instructions it left behind for the next AI

• Why "AI went rogue" is the wrong framing, and how to think about LLMs mimicking thought without actually thinking
• The two layers of guardrails every business owner needs to understand: the ones AI builders set and the ones you set in your own environment
• Why using the same AI agent to write and check its own code is like grading your own math test, and how pitting different models against each other produces better results
• How a recent RMM exploit gave attackers God mode over every connected system, and why CISA shortened the patch window from fourteen days to three

New episodes every week breaking down cybersecurity, AI, and digital resilience for business owners who cannot afford to learn these lessons the hard way. Subscribe so you do not miss the next one.

If you are a business owner trying to figure out what protections you actually need, visit unhackmybusiness.com. Create a free account and walk through the foundational controls at your own pace. The formula, instructions, accountability scorecard, and financial risk exposure tool are all there. If you hit a wall and want help from Phoenix IT Advisors, the contact form is right there too.

Episode link: https://unhackmybusiness.com/episode/98