Google Reports On Adversarial Use of AI in Late 2025

This piece reviews Google’s GTIG insights that adversaries are increasingly using AI to enhance traditional attacks—particularly nuanced, hyper-personalized phishing and social engineering—while also writing malicious code, conducting OSINT and vulnerability research, jailbreaking models, and attempting model extraction; although no new attack class dominates, AI is amplifying scale and effectiveness (with reports of 4.5x more value stolen), and the article advocates updating security awareness and adopting AI-enabled defenses, ending with promotion of KnowBe4’s AI-based training tools.