Tomato Novel (番茄小说) purges AI-generated "low‑quality" web novels after surge of Star‑Moon Writing titles
What happened
Tomato Novel (番茄小说) has pulled dozens of high‑profile titles it says were produced with low‑quality AI assistance, igniting renewed debate in China’s online fiction community about the limits of automated writing. The platform’s May notice on tightening low‑quality content governance explicitly bans “bulk generation by abusing AI tools” and says it has refused to sign 112,700 low‑quality books and removed more than 40,000 problematic titles. Among the most visible casualties were multiple bestsellers promoted by Star‑Moon Writing (星月写作), an AI‑assisted writing service from Zhiyu Technology (智语科技).
One removed novel, titled 重生成树，开局妹妹把我上交国家, had reportedly topped reading charts with over one million in‑reader counts before being taken down; the official reason given was “AI low‑quality expression” rather than plagiarism. Other Star‑Moon‑promoted works — including 国运求生：我靠签到万倍返还龙国 and 开局拍卖寿元，圣王大帝抢疯了 — were also delisted. It has been reported that Star‑Moon’s small founding team scaled early growth by showcasing income cases and chart successes to attract paying users; claims of large creator earnings and platform traffic have become a flashpoint in the controversy.
Platform response and industry context
Tomato Novel has followed earlier moves by other major players to curb mass‑produced AI output. In February it flagged misuse of AI and bulk updates as priorities for enforcement; on May 21 it introduced new posting limits per author and per work, and activated facial‑recognition binding for withdrawals to prevent studio‑style mass account abuse. Qidian (起点中文网) and other incumbents have grappled with similar complaints for years, as authors and readers complain that templated AI novels crowd out original work and degrade reader experience.
Why the fuss now? Because automated tools can scale rapidly and game discovery systems, sometimes creating the appearance of popular titles through coordinated uploads and dubious engagement. Some observers have suggested that chart wins for AI‑assisted works were inflated by botting or mass production rather than organic readership — it has been reported that in many cases works showing hundreds of thousands “in‑read” had only small daily active催更 counts. That has made moderators and platforms warier, and it has raised broader questions about how to police algorithmic content generation at scale.
Why it matters
The takedown underscores a broader tension in China’s tech ecosystem: platforms want to monetise AI innovation, but they also face pressure to protect creators’ livelihoods and platform reputations. For readers, the purge promises cleaner recommendation feeds. For small authors, it could restore exposure and income; for AI toolmakers, it signals that rapid monetisation without robust quality controls may invite swift pushback. Can platforms effectively distinguish bad‑faith mass production from legitimate AI‑assisted creativity? The answer will shape China’s enormous web‑fiction market and offer lessons for regulators and platforms worldwide as AI content generation proliferates.