ByteDance, Alibaba to shut down personalized AI agent services amid tightening regulations

Chinese tech giants ByteDance and Alibaba will shut down personalized AI agent services on their respective platforms — Doubao and Qwen — later this month, as Beijing tightens safety regulations governing the technology.

China's state-run Global Times and other outlets reported Monday that Doubao and Qwen announced they would discontinue their personalized AI agent features starting July 15.

Unlike standard AI chatbots that simply answer questions, personalized AI agents are designed to remember a user's preferences and conversation history and carry out complex tasks on the user's behalf.

ByteDance said Doubao would retain user data for a set period after the service ends, during which users could review agent information and conversation logs or download them. China Securities Times said the preserved data would be deleted about three months later.

The announcements drew backlash on Chinese social media platforms including Weibo, with some users voicing opposition to the discontinuation of services they had long relied on, the outlets said.

Global Times said the shutdowns may reflect a combination of factors, including the need to comply with China's tightening AI safety requirements and a pullback from investment in projects with limited commercial viability.

The Chinese government announced interim regulations on the management of AI anthropomorphic interaction services — commonly known as personalized AI agents — in April, with enforcement set to begin July 15.

The rules impose strict requirements on platforms, including the establishment of anti-addiction systems, identity verification for minors and rigorous content screening.

Separately, guidelines the government issued in May to promote the development and application of AI agents defined for the first time the boundaries of decision-making authority between AI agents and users. The guidelines specify that users must retain the ultimate right to be informed of and to override any decisions made by an AI agent.

Global Times said China's AI agent industry has entered a phase of rapid expansion, but that violations have emerged on some platforms — including fake agents impersonating official institutions, borderline services offering crude or extreme role-play conversations, and security risks involving the unauthorized collection of personal data.

Authorities have moved to address the problem. The Shanghai Cyberspace Administration deleted more than 14,000 non-compliant AI agents last month.