Large Language Models: Moving Past the Early Stage – Source: www.databreachtoday.com

Artificial intelligence, machine learning and large language models are not new, but they
are coming to fruition with the mass adoption of generative AI. For cybersecurity professionals, these are “exciting times we live in,” said Dan Grosu, chief technology officer and chief information security officer at Information Security Media Group.

Enterprises have always collected a lot of data, but they haven’t always done a good job of gaining insights from it. Now, with generative AI, companies have the opportunity to use the data they have to maximize their security. But, Grosu said, “Even with fine-tuning of these LLMs, we are still very far from addressing ethical concerns, biases, fairness and responsible usage of these models – specifically, in a cybersecurity context, where the risks are huge.”

In this episode of CyberEd.io‘s podcast series “Cybersecurity Insights,” Grosu discussed:

• The need for guardrails in the use of LLMs and the importance of giving them carefully considered instructions;
• If and when LLMs will ever achieve true sentience;
• Whether cybersecurity practitioners in businesses can learn from the cybersecurity strategies used by nation-states.

As a leader in technology and security at Information Security Media Group Corp for the past 14 years, Grosu has been pivotal in building and managing the company’s infrastructure from its inception, guiding its successful transition to cloud computing. He is also the technical co-founder of CyberEd.io. Grosu has over 20 years of experience in shaping technology strategies for startups and SMBs. Prior to ISMG, Grosu held key roles in organizations across banking, healthcare and automotive sectors.