Chinese chatbots drop AI personas under new Beijing rules

Chinese tech giants pull personalised chatbot characters

Several of China’s largest technology companies have announced plans to remove AI persona features from their flagship chatbots, aligning their products with a tighter regulatory framework being imposed by Beijing. Nikkei Asia reported that Tencent, Alibaba and ByteDance are all moving to comply with the new rules, and identified NetEase as another company pulling such personalised features from its products. The Nikkei report did not include a date for the change or specify which chatbots are losing which characters, but the breadth of the companies affected suggests a sector-wide adjustment rather than a single-product fix.

What are AI personas and where were they used?

AI personas, sometimes called “AI companions” or character chatbots, are configurable digital personalities that allow a generative-AI system to adopt a specific tone, backstory or name. They became a popular design layer for Chinese consumer chatbots in 2024 and 2025, with services marketed around fictional friends, historical figures or cartoon avatars designed to keep users chatting for longer. The Nikkei report did not list specific persona products by name, but did describe the change as the disabling of an entire category of personalised features rather than the deletion of a single character.

The regulatory backdrop

The removal drive comes as China tightens its framework for the rapidly advancing AI industry. Nikkei described the move as part of Beijing’s broader crackdown-style tightening, though the article excerpt did not quote the specific regulation, measure number or issuing agency. The Cyberspace Administration of China (CAC), the country’s main internet regulator and the body that has overseen prior generative-AI rules, was not named in the supplied source material. Readers should therefore treat the precise legal instrument as unconfirmed from the sources provided, even if the broader trajectory of stricter Chinese AI oversight is well documented.

Which companies and products are affected

According to Nikkei Asia, the companies named so far include three of China’s biggest consumer-tech platforms — Tencent, which operates the Yuanbao chatbot; Alibaba, which runs the Tongyi and Qwen-backed assistant; and ByteDance, whose Doubao chatbot has become one of the most-used AI assistants in China. NetEase was also identified by Nikkei as among those pulling such personalised features, consistent with its mix of gaming and consumer-internet products where character-driven AI is a natural fit. The article did not break down the timeline or scope of changes at each company, leaving the operational specifics for each platform to be confirmed in the firms’ own statements.

How the changes compare with existing Chinese AI rules

China has, since mid-2023, required generative-AI services to obtain a licence from the CAC and to follow rules on training data, user data and content. The current persona crackdown appears to extend that oversight into the design layer of products, where chatbots are explicitly given a name, voice or identity. The Nikkei report treats this as a tightening rather than a wholly new regime, but does not cite a fresh legal document in the excerpt reviewed, and does not specify whether the changes are mandatory or voluntarily pre-emptive on the part of the platforms. That distinction matters: pre-emptive removals can be reversed, while regulator-ordered bans typically come with penalties.

Sourcing and what remains unverified

Only one source in the supplied set — the Nikkei Asia report — directly addresses the chatbot-persona story. The second source, a Deutsche Welle feature on China’s rise in science and technology, was supplied for context. It frames AI as one of the “key future industries” in China’s draft 15th Five-Year Plan for 2026–2030, alongside quantum technology, nuclear fusion, biotechnology, brain-computer interfaces and deep-sea and space exploration. That framing supports the broader narrative that Beijing considers generative AI strategically important and therefore politically sensitive, but the DW article does not mention the persona crackdown directly. Readers should treat any specific numbers or quotes on the policy itself as drawn from Nikkei, not from the DW piece.

Stakes for Chinese consumers and the global market

Chinese chatbots from Alibaba, ByteDance and Tencent each count users in the hundreds of millions, and Doubao in particular has been cited in industry coverage as among the world’s most-used AI assistants. Removing persona features is likely to change the user experience across those products and could affect engagement metrics on which consumer AI services are increasingly judged. Beyond China, the move matters because Chinese large-language-model providers — including Hangzhou-based DeepSeek and others — have been exporting model weights, APIs and chat products into markets from Southeast Asia to the Middle East. A regulatory shift that alters Chinese products could push Chinese firms toward more sanitised exports or, conversely, accelerate the spread of personality-rich alternatives from non-Chinese providers.

Differences in tone between the two sources

The two supplied sources approach the Chinese tech sector from different angles. Nikkei’s reporting is granular and business-focused, naming specific firms and treating the persona change as a discrete regulatory event. The DW feature is broader and more thematic, treating China’s technological ascent as part of a long-running competition with the United States and citing the Nature Index — which puts China first in research leadership — and the views of Germany’s Max Planck Society and the German Research Foundation. For this story, only the Nikkei material is on-point; the DW context is included to anchor the regulation within the larger pattern of Beijing treating AI as a strategic industry, but it does not add new facts about the chatbot policy itself.

Why it matters beyond China

Generative-AI services operate across borders even when their operators do not. Chinese chatbots’ persona features have been studied as design templates by product teams in the United States, Europe and elsewhere, and several Western startups have shipped character-driven chatbots modelled in part on Chinese competitors. If Beijing’s rules force a wholesale rethink of personas on the Chinese side, design choices elsewhere — ranging from voice style to memory features — could be reassessed, either preemptively to avoid future regulatory friction or because Chinese-developed open-weight models lose a distinctive feature. The Nikkei report does not address these downstream effects, but they are a reasonable inference from the scale of the platforms involved.

What to watch next

Key milestones in the coming weeks will include any formal statement from the CAC clarifying whether the changes are mandatory, a deadline by which platforms must comply, and any list of persona categories that are now off-limits. Watch too for product-level announcements: Yuanbao, Tongyi, Doubao and NetEase’s consumer chatbots have historically published release notes when features are added or removed. On the policy side, the draft 15th Five-Year Plan, which DW reports highlights AI among its priority industries for 2026–2030, is a likely vehicle for further formalisation of these rules. International readers should also monitor how rival chatbot providers — Baidu’s Ernie Bot, Moonshot’s Kimi, DeepSeek’s assistant and MiniMax-M3-powered services — respond, because uniform adoption would suggest a regulatory directive, while uneven adoption would point to voluntary risk management.