Shifting Cybersecurity: The Impact and Implications of LLMsProofpoint CEO Ashan Willy on Taking an Experimental Approach to Applying Gen AI
While a significant number of attacks are not yet AI-driven, there's a noticeable shift in the creation of generative malware and lures for business email compromise, warned Ashan Willy, CEO at Proofpoint. Large language models are being used to create enticing lures in foreign languages to target broader audiences.
Using LLMs for defense is challenging due to their high cost and slow implementation across the attack chain, Willy said. "LLM is anywhere from five to eight times more expensive than general AI to use," he said. "So where you use it in the pipeline is important."
LLMs can summarize documents and classify data. They have the potential to revolutionize user interfaces by transforming the way security analysts query information, Willy said. He predicts user interfaces will undergo a significant transformation over the next few years due to the impact of generative AI.
The AI conversation is in its early stages, akin to the early days of the internet. It is necessary to remain experimental and open-minded in exploring the potential applications and implications of LLMs and generative AI. Any definitive conclusions about AI's future are premature, he said.
In this video interview with Information Security Media Group at Black Hat USA 2023, Willy also discussed:
- Using LLMs for email classification and sentiment analysis;
- How Proofpoint helps customers comprehend the implications of LLMs and generative AI;
- How generative AI can be used to create malware and customize cyberattacks.
At Proofpoint, Willy directs a global workforce of over 4,000 employees focused on defending organizations from advanced threats and compliance risks. He has more than 25 years of leadership experience in sales, channel development, engineering, product management, customer success and worldwide go-to-market strategy across global markets.