AI Faces Perceived More Trustworthy Than Real - 조선일보

A joint study by Lancaster University, Stanford University, and UC Berkeley reveals that AI-generated faces are perceived as more trustworthy than real human faces. Participants struggled to distingui... Read original →

The 'trustworthiness paradox' exposed by this research will force a full-stack overhaul of identity verification systems. Upstream chipmakers must accelerate integration of dedicated AI detection IP blocks—especially real-time deepfake classifiers in edge SoCs. Downstream, social media and fintech firms face surging compliance costs as regulations like the EU’s DSA and China’s interim generative AI rules mandate synthetic content labeling, yet current watermarking schemes remain vulnerable to fine-tuned diffusion models. NVIDIA and AMD are racing to embed detection inference units into AI accelerators, while Qualcomm bets on lightweight on-device discriminators to reduce cloud dependency. Within 18 months, deepfake-enabled fraud will catalyze global KYC standard upgrades, merging trusted computing with privacy-enhancing technologies. Crucially, as 'averaged faces' become prime vectors for deception, biometrics must shift from static features to dynamic behavioral modeling—a paradigm shift not just in algorithms, but in semiconductor-based perception architecture itself.