Douyin’s senior executive has signalled a deliberate pullback from the freewheeling era of generative-video demos. On February 15, Li Liang, a vice‑president at Douyin Group, urged users to try the new Seedance 2.0 features — branded inside Douyin as Doubao and Jimeng — but stressed that generating a digital double requires real‑person verification and that the platform currently blocks face references and the creation of recognised IP characters such as Disney properties or popular Chinese cartoons.
The announcement is striking less for what Seedance can do than for what Douyin will not allow. Behind the user‑facing limits lies a heavier investment: Li said the team’s biggest recent effort has been continuously maintaining and strengthening anti‑infringement strategies, and invited users to flag problematic outputs for removal. That message underscores how quickly content moderation, copyright risk and regulatory compliance have become operational priorities for Chinese platforms rolling out sophisticated AI content tools.
Seedance 2.0 arrives into a crowded and fraught landscape. Chinese companies have pushed a wave of multimodal generative models and consumer apps in recent months; some integrations have sparked fanfare as well as legal headaches when AI imagery used celebrities’ likenesses or copied trademarked characters. Regulators in China and abroad are also sharpening rules on deepfakes, privacy and intellectual property, while rights holders increasingly demand controls or licences to prevent brand dilution and unauthorised exploitation.
The requirement for real‑person verification is an obvious compliance tactic: it helps Douyin identify who is creating synthetic likenesses and therefore makes takedown and accountability mechanisms more practicable. Yet it also raises tradeoffs. Identity checks put an extra friction point between users and creative experimentation and could deter casual creators or small studios. At the same time, refusing to permit “true‑to‑life” facial references or IP character generation limits the fidelity of the product and may push specialist users toward off‑platform tools or illicit workarounds.
For rights holders, the move is reassuring; for regulators, it is evidence that a major platform is proactively managing risk. For ByteDance and its peers, the calculus is to accelerate innovation without courting high‑cost litigation, regulatory intervention or reputational damage. How companies balance product attractiveness with enforcement and licensing will determine whether these generative platforms scale sustainably or become a series of sandboxed features with limited commercial reach.