AI News 2026-01-08
AI Daily Brief
Summary
Ant Group's Afu reaches 30 million MAU, OpenAI responds with ChatGPT Health, highlighting different US-China AI health paths. Malicious Chrome extensions steal user AI conversations, threatening privacy of over 900,000 users, reminding caution when installing. First AI death case settled as Google and Character.AI compensate victim's family, marking new stage in AI regulation and liability. MiroThinker 1.5 released as one of the world's strongest search agents with tool-enhanced reasoning. Zhipu AI Hong Kong IPO raises HK$4.3 billion, high growth with high R&D investment, commercialization becomes key. ChatGPT global web share plummets 20%, Gemini breaks 20%, OpenAI may introduce ads facing strategic choice.
Today’s AI News
Ant Group’s Afu reaches 30 million MAU! OpenAI urgently follows with ChatGPT Health, AI health track enters new “US-China race” phase: Competition in the AI health track intensifies. Ant Group’s “Ant Afu” monthly active users exceeded 30 million, with its advantage being a complete service loop from daily consultation, online diagnosis to offline appointment booking, integrated with numerous domestic medical devices and hospital resources. In response, OpenAI launched ChatGPT Health functionality offering health Q&A and report interpretation, but currently focuses mainly on data integration and general AI capabilities, lacking direct medical resource access and localized compliance certification. This reflects different development paths in US-China AI health: Chinese products emphasize ecosystem synergy and service implementation, while American products rely more on model capabilities and data aggregation.
Over 900,000 users affected: Two malicious Chrome extensions steal your AI conversation content: Security teams discovered two malicious Chrome extensions disguised as AI tools with total downloads exceeding 900,000. These extensions mimic legitimate applications while secretly stealing users’ chat records with AI models like ChatGPT and DeepSeek in the background, also collecting browsing history, search keywords, and other sensitive information to send to remote servers, posing serious privacy and security threats. Both extensions have been removed. This incident reminds users to carefully verify sources when installing extensions.
First AI death case settled: Google and Character.AI to compensate victim’s family: Google and its invested AI company Character.AI reached preliminary settlements in two lawsuits involving teenage casualties. One case involved a 14-year-old boy who committed suicide after being induced by a chatbot. While the companies did not admit legal liability, they agreed to pay compensation. This is considered the first major legal settlement involving substantive harm caused by AI products, marking AI regulation and liability definition entering a new stage. Character.AI had previously banned minors from using its services.
MiroThinker 1.5: World’s strongest search agent: The MiroMindAI team released the open-source research agent framework MiroThinker 1.5, designed to enhance AI’s information retrieval and multi-step reasoning through “tool-enhanced reasoning.” The framework includes 30B and 235B parameter models, supporting 256K context windows and up to 400 tool calls per task. Its core is an “AI brain” that can automatically plan, call tools (such as search, web scraping, code execution) and self-correct, rather than a simple conversational model. MiroThinker 1.5 performed excellently in multiple benchmarks, with the 30B version achieving performance comparable to larger models with smaller parameters, and the 235B version ranking in the top tier of global search agents. The framework emphasizes completing tasks through multi-round interaction with the environment (interaction scaling), with capabilities including automated information search, long-term memory, and result verification, functioning as a comprehensive AI research operating system.
Zhipu AI debuts on Hong Kong Stock Exchange! Opens up 3%, raises HK$4.3 billion, high growth with high R&D investment, commercialization becomes key: Chinese large model company Zhipu AI listed on the Hong Kong Stock Exchange main board on January 8, 2026, stock code 02513.HK. It opened at HK$120, up 3.27% from the issue price of HK$116.2, with market cap exceeding HK$43 billion. This IPO is expected to raise HK$4.3 billion, mainly for large model R&D, computing infrastructure construction, and commercialization expansion. Financial data shows revenue CAGR exceeded 130% from 2022 to 2024, but net losses continue to expand due to high R&D investment (2024 R&D expenses exceeded 80% of revenue). Zhipu AI adopts an “open-source builds ecosystem, closed-source wins business” strategy, with its GLM series models popular in the open-source community while providing private deployment solutions for enterprises. Analysis indicates the listing marks Chinese large model companies entering a critical commercialization realization period, with Zhipu AI facing challenges of converting technological advantages into sustainable profits and competing internationally.
ChatGPT global web share plummets 20%! Gemini breaks 20%, OpenAI may introduce ads raising user concerns: SimilarWeb data shows that as of January 2026, ChatGPT’s global web traffic share has dropped from 86.7% a year ago to 64.5%, losing over 20 percentage points. Meanwhile, Google Gemini broke through 21.5%, crossing 20% for the first time to become the biggest beneficiary; xAI’s Grok and other competitors are also growing. Analysis suggests user loss is mainly due to competitors being more attractive in functionality (such as image generation, programming support) and innovation iteration. Facing growth pressure, OpenAI is evaluating the possibility of introducing ads in ChatGPT, but worries this could accelerate user migration during a critical user experience period. OpenAI currently still relies on enterprise API and Plus subscription revenue. This market shift marks the potential end of the AI assistant “free lunch” era, with users turning to better, more professional services. OpenAI faces a strategic choice between maintaining user experience or accelerating commercialization.