Bilibili hires Anuttacon researcher as first AI video-generation lead

Bilibili has named its first dedicated head of AI video generation, according to 36Kr, hiring a researcher whose last job was inside an AI lab set up by the founder of the game studio Mihoyo. The move gives the Chinese long-form video platform a named owner for the production side of its AI stack at a moment it has spent two years building out creator-side tools.

Bilibili has not responded to a request for comment on the report, which 36Kr published Friday and attributes to people familiar with the hire. The new lead, Zeng Ailing, will report directly to CEO Chen Rui.

Zeng joins from Anuttacon, a small AI startup founded in 2024 by Cai Haoyu, the creator of the Mihoyo game franchise Genshin Impact. At Anuttacon she led work on human-centered, interactive, multimodal video generation, a focus that reads as closer to a creator-tooling product than to a recommender system. She holds a PhD in computer science from the Chinese University of Hong Kong and was a visiting scholar at Carnegie Mellon's Robotics Institute.

Her brief is the production stack. Bilibili's public framing of its AI investment has, until now, been video understanding, video recommendation, and assisted video creation. The first two are consumption-side: they decide what appears in a user's feed. The third is the thin edge of the production side: tools that help a creator edit, dub, or auto-cut a video they have already shot. A dedicated video-generation lead, with a background in interactive multimodal video, sharpens that third leg into its own org. It is the first time the chart has caught up to the framing.

The structural claim is falsifiable. Bilibili already shipped two creator-side AI products without a named owner for the function. In March 2026 the company launched updream, an AI video creation platform built around long-term memory, a personalized skill library, and a free canvas for assembling clips. In July 2026 it shipped Toy, a browser-based publishing tool that turns a local interactive web page into a playable post on Bilibili; during the June-July beta, top creators uploaded more than 300 works and Toy logged over 8.5 million user interactions. Both products live on the production side. Neither was attributed, at launch, to a specific AI lead.

Open-source work at Bilibili points the same direction. The company has published Index-1.9B, a lightweight multilingual language model, and Index-anisora, an open-source anime video generation model whose paper is on arXiv. AniSora is an open-source anime video generation model Bilibili published on arXiv. A dedicated head does not change that, but it does give the next version of those products a person to attach to.

The gap behind the hire is not closed by filling it. The 36Kr report flags that Bilibili has been searching for a stable AI lead since September 2025, when the prior holder of the role, a former Kuaishou vice president, departed; 36Kr's published excerpt is truncated and does not name that predecessor. The piece is therefore most useful read as "the company is trying again, with a creator-tooling profile," not as a settled appointment. The earlier leader's name and the reasons for leaving are the kind of detail a confirmed report would carry, and this one does not.

Chen Rui, Bilibili's CEO, made a separate public statement on the Q1 2026 earnings call, framing AI as a "10x amplified historic opportunity" that did not change Bilibili's underlying logic but magnified its two existing advantages. The line is worth pinning to its source, because it is a year-old, board-level read on the platform rather than a comment on the Zeng hire. The two questions to watch are whether Bilibili ships a creator-facing video generation product in the next two quarters that visibly carries her direction, and whether the prior leader's gap ever gets a name.

The company has also been using AI to advertise. In its most recent quarter, Bilibili said more than 190 million users per month watched AI-tagged content on the platform, that AIGC-made creative assets now account for more than half of its ad spend, and that automated creative tools have penetrated more than 85 percent of ad consumption. Those are Bilibili-side numbers, not independent measurements, and they describe the ad business, not the video-generation business Zeng is being hired to run. The ad numbers are also the reason a dedicated head matters: the production stack is the part of the AI budget that is starting to show up in revenue lines, even before the creator products do.