AI Videos Flooding the Internet: Users Need to Strengthen Their Discernment

When scrolling through short videos, have you ever doubted that the people you see might not even be real? Recently, on platforms like Douyin, Xiaolei noticed comments under some abstract or funny videos: “Is this AI?”

The rise of AI-generated videos and images has made people instinctively question the authenticity of what they see. Although platforms are strengthening supervision and users are more cautious, the “tricky moments” of AI videos are just beginning.

For instance, a few months ago, deep-night live streams featured “celebrity endorsements” by Zhang Wenhong, Tony Leung, and Andy Lau. In reality, these videos may have been entirely AI-generated. Creating virtual humans has become extremely accessible: simply upload a person’s facial footage and combine it with AI voice synthesis to create a highly realistic digital human, without their participation. Some businesses target elderly users, using virtual celebrities to promote low-quality, overpriced products.

Globally, AI video regulation faces challenges. TikTok’s “Homeless Challenge” caused parental panic and even police reports, showing that the fast spread and realism of AI videos can pressure public opinion and consume public resources.

AI videos also harm original content creators. Low-cost AI-generated videos flood platforms, causing content homogenization and burying high-quality originals. While platforms like Douyin have carried out targeted regulation, funny or entertaining AI videos still easily gain traction.

Key ways to spot AI videos include:

• Abnormal backgrounds and lighting;
• Overly smooth or symmetrical faces;
• Repetitive, unnatural movements;
• Failure to respond to simple interactive instructions in live streams.

The most effective defense is maintaining a critical mindset and intuition: if a video feels off, ask “why?” at least a few times. Meanwhile, tech companies, platforms, and regulators should work together, for example, embedding digital watermarks to label AI content and upgrading detection mechanisms, using AI to fight AI.

In today’s era of increasingly realistic AI videos, staying vigilant and sharpening your discernment remains the best safeguard.