Adversarial Intelligence: Red Teaming Malicious Use Cases for AI

Recorded Future’s Insikt Group assesses that in 2024 the most likely malicious uses of AI will be targeted deepfakes and influence operations, with open-source tools lowering costs to impersonate executives, clone websites, and fabricate media. The research finds AI can also aid malware evasion through code alterations and support reconnaissance via multimodal analysis, though human oversight remains key for believable spoofs and actionable intelligence. Organizations should expand their attack surface management to include executive voices and likenesses, branding, and public imagery, and prepare for stealthier detection challenges as threat actors increasingly adopt generative AI.