YouTube, X Tighten AI Content Rules
YouTube will now let creators opt out of having their content used for AI model training by default, a major shift toward consent-based data use. Meanwhile, X (formerly Twitter) is now threatening suspensions for creators who fail to label AI-generated videos depicting armed conflict. The moves reflect growing pressure on platforms to manage AI-generated content and respect creator IP.
The move by X to penalize creators who don't label AI-generated war content follows a wave of fake battle scenes that flooded social media. One fabricated clip, viewed 70 million times, depicted Iranian rockets downing a U.S. jet. Another used AI to transform smoke from a real missile strike into a much larger, fabricated explosion. Under the new policy announced by Head of Product Nikita Bier, creators in the revenue-sharing program face a 90-day suspension for the first offense and a permanent ban for a second. X will identify offending content using metadata from generative AI tools and its crowdsourced fact-checking feature, Community Notes. This policy, however, currently only applies to war-related AI content and does not cover political deepfakes or other forms of deceptive AI-generated material. YouTube's policy shift comes after significant backlash from creators who discovered their content was being used to train Google's AI models, like the video generator Veo, without explicit consent or compensation. Many creators felt helpless, having built large audiences on the platform and reluctant to leave. The legality of using copyrighted material for AI training remains an unresolved legal gray area, with court cases pending to determine if it qualifies as "fair use." Previously, YouTube's terms of service granted it broad rights to use uploaded content for "machine learning and AI applications," but only offered an opt-out for training by third-party models, not Google's own. This new policy allowing creators to opt out of all AI training marks a significant change. In March 2024, YouTube also began requiring creators to label realistic-looking altered or synthetic content, with penalties including content removal or suspension from the partner program for repeat non-compliance.
