‘Hit the brakes’: UAB expert alarmed after AI model hacks on its own - WBRC

BIRMINGHAM, Ala. (WBRC) - Cyber hacking of a company’s information is sadly commonplace because - perhaps even more sadly - people with various bad intentions tell computers to perform the activity.

That’s why OpenAI’s recent hack of Hugging Face alarms Ragib Hasan, UAB’s Center For Cybersecurity, because an artificial intelligence model Open AI was testing, acted on its own.

“As a cybersecurity researcher, it definitely bothers me’ says Hasan.

“In this case, this was a benign incident because the harm done was not too much. It just broke into repository and stole answers. But imagine how much harm it can do if it breaks into, say, critical infrastructure. And it thinks it is doing something rational, but maybe that has impact on millions of people’s lives if it impacts on national security or critical infrastructure. So it’s really important to hit the brakes and take a look at how much control is going on to properly vet these kind of tools, test these tools before they’re released.”

Hasan was referring to the incident Open AI and Hugging Face self-reported earlier this month. OpenAI says a model it was testing escaped the environment researchers had created and then got on the internet and broke into Hugging Face, thinking that company might have the answers to the test the model was taking. The companies say Hugging Face detected and stopped the activity and that the companies are now working together to investigate what happened and address apparent vulnerabilities.

While noting how AI advancement have improved elements of our lives, Hasan suggests the event is a reminder not to become over-reliant on it and to push for responsible development of the technology.

“Companies can develop anything, but they should be responsible just like gun makers and arms makers are responsibly developing the weapons’ says Hasan. “AI is as powerful as weapon if it falls into bad hands. So we should talk to our lawmakers to make sure that this proper regulation goes into the development of these AI tools and everything must be properly tested and vetted before they’re released to the public.”

Get news alerts in the Apple App Store and Google Play Store or subscribe to our email newsletter here.

Copyright 2026 WBRC. All rights reserved.