Microsoft warns deepfake election subversion is disturbingly easy

Microsoft’s Threat Analysis Center GM Clint Watts outlines how state-linked actors are leveraging low-effort AI-enabled influence tactics ahead of 2024 elections, noting that simple manipulations—like images overlaid with real news logos and especially AI-generated audio lacking contextual cues—are proving highly effective. He traces the evolution of information operations from early troll activity to today’s video-centric platforms and offers heuristics for detection, emphasizing that content set in private settings and blends of real and fake audio are the hardest to verify.