China airs first long-form AI-generated series

Summary

China has released 'Beyond Wukong,' a 30-episode animated series with 40-minute episodes, generated end-to-end using ByteDance's Seedance 2.0/2.5 video models. The series premiered in late 2026 on Mango TV and Hunan Satellite TV. Reports indicate a production cost of approximately 900,000 yuan per episode, with the process taking five months. The director claims these figures represent one-tenth of traditional animation costs.

Why it matters

This production is notable as the first long-form AI-generated series to air during prime-time broadcast hours. Unlike short promotional videos, it serves as a genuine pilot study on the economic sustainability of AI-native entertainment content. China's regulatory authority has explicitly integrated AI into the content approval process, allowing content production and censorship review to proceed in parallel. This points to a structural policy shift unmatched by any Western market. The contrast with AI-related labor disputes in Hollywood highlights the deepening global divergence in the industry's approach to generative tools.

Highlights

• 30 episodes of 40 minutes each, produced in ~5 months by a team of approximately 100 on Mango TV's internal platform
• ~€116,000 per episode; the director claims it is 10 times cheaper and 70% faster than traditional animation
• China's audiovisual regulatory authority accepts AI as an explicit part of content policy, enabling simultaneous production and approval processes

Critical take

The article presents claims of a 10x cost advantage and 70% speed increase as fact without offering independent verification. While a peak TV rating of 0.34% is framed as a success, it remains modest by any broadcasting standard. This suggests the real story may lie in regulatory signaling rather than audience demand. The text also equates the phrase 'first AI series on TV' with broad-scale quality or artistic achievement, failing to provide any critical assessment of the actual visual or narrative output.