Unverified50% confidenceFactExact time
The technical pipeline for digital human video generation is now relatively mature, but generation quality varies widely, and details such as teeth and eye expression still easily reveal AI artifacts
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/7/2026
First Seen
Valid until: 1/5/2027
Sources
Related Entities
Related Claims
Unverified随着新一代视频生成模型演进，遮挡识破假脸的破绽窗口期可能并不长久78% similarUnverifiedThe realism of synthetic media has approached or even surpassed the limits of human visual identification, making it difficult to distinguish AI-generated content from real video.75% similarUnverified当前许多平台的人脸识别系统仍然依赖2D视频验证，容易被AI换脸技术突破73% similarUnverifiedThe hardest part of video face-swapping is not individual frames, but the temporal consistency of facial expressions, head rotation, and lighting changes between consecutive frames70% similarUnverifiedOver the past one to two years, generative video and image models have made noticeable progress in detail, lighting consistency, and temporal coherence, reducing the flickering and distortion problems previously seen in AI visual effects.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/985211API
curl https://kongchang.com/api/v1/knowledge/claims/985211MCP
get_claim(id=985211)