![[6d4be853-58c8-40f9-ac81-653009842f9b.mp3]]

苹果AI发展受质疑，WWDC成检验信心关键时刻

简报：

• 苹果此前高调承诺为iPhone引入一系列AI功能，至今大多未兑现，导致外界质疑其在AI竞赛中落后于Google、OpenAI、三星等竞争对手。
• 即将召开的WWDC被普遍视为苹果“挽回信任”的转折点，预计将公布与更多AI合作方（如Google、Perplexity）的新进展，并可能首次向第三方开发者开放AI基础模型以刺激创新。
• 分析认为，Siri智能升级继续推迟、开发者社区对苹果抽佣及生态封闭表达不满，内部（如Jony Ive与OpenAI的合作）和外部（美国关税威胁）挑战亦加剧苹果压力。
• 业界关注苹果能否借WWDC及时重振AI信心、缩小与主要科技巨头的AI差距，或将决定其在新一轮科技革新的话语权。

相关链接：

• iPhone神话动摇？苹果在AI竞赛中步履蹒跚，WWDC或成关键转折点
• 彭博：蘋果WWDC聚焦新操作系统与AI合作
• 苹果AI布局挤压，内部研究质疑现有AI推理能力

Apple's AI Development Under Scrutiny, WWDC a Key Moment to Restore Confidence 🧐

Brief:

• Apple's previous high-profile promises to introduce a series of AI features to the iPhone have largely gone unfulfilled, leading to external doubts about its lagging position in the AI race compared to competitors like Google, OpenAI, and Samsung. 📉
• The upcoming WWDC is widely seen as a turning point for Apple to 'regain trust.' It's expected to announce new progress with more AI partners (e.g., Google, Perplexity) and potentially open its foundational AI models to third-party developers for the first time to stimulate innovation. ✨
• Analysts suggest that the continued postponement of Siri's intelligent upgrade, coupled with developer community dissatisfaction regarding Apple's commission fees and closed ecosystem, and internal (e.g., Jony Ive's collaboration with OpenAI) and external challenges (e.g., US tariff threats) are exacerbating pressure on Apple.
• The industry is watching closely to see if Apple can leverage WWDC to timely reinvigorate AI confidence and narrow the AI gap with major tech giants, which could determine its influence in the new wave of technological innovation.

Related Links:

• Is the iPhone Myth Shaking? Apple Stumbles in the AI Race, WWDC Could Be a Key Turning Point
• Bloomberg: Apple's WWDC to Focus on New Operating Systems and AI Partnerships
• Apple's AI Layout Under Pressure, Internal Research Questions Existing AI Reasoning Capabilities

通义千问3模型开源，全球下载量突破1250万

简报：

• 阿里通义千问3大模型自开源一个月以来，全球累计下载量已超过1250万，在Hugging Face、魔搭社区、Ollama等主流AI开源平台引发广泛关注。
• 该系列已开放0.6B、8B、30B和32B四个模型尺寸，各版本下载量均突破百万，显示出强劲的全球开发者吸引力。
• 在Hugging Face平台，通义千问3的衍生模型数量已超过13万个，跃居全球第一，反映出开发者社区的高度活跃与创新热情。
• 通义千问系列因开源策略透明、性能优异和部署灵活，已成为全球开源AI模型生态中的热门，并在代码助手、智能客服、Agent开发等领域获得广泛应用。
• 行业分析认为，千问3的爆发式增长显示国产大模型开源实力正与国际接轨，并进一步增强了阿里在全球AI基础模型生态中的影响力。

相关链接：

• 通义千问3大模型全球爆火：下载量超1250万
• 阿里通义千问3模型开源首月下载量破1250万

Tongyi Qianwen-3 Model Open-Sourced, Global Downloads Exceed 12.5 Million 🌍

Brief:

• Since its open-sourcing one month ago, Alibaba's Tongyi Qianwen-3 large model has accumulated over 12.5 million downloads globally, drawing significant attention on major AI open-source platforms like Hugging Face, ModelScope, and Ollama.
• The series has released four model sizes: 0.6B, 8B, 30B, and 32B. Each version has surpassed one million downloads, demonstrating strong global developer appeal.
• On the Hugging Face platform, the number of derivative models based on Tongyi Qianwen-3 has exceeded 130,000, ranking first globally. This reflects the high activity and innovative enthusiasm of the developer community. ✨
• The Tongyi Qianwen series has become a hot topic in the global open-source AI model ecosystem due to its transparent open-source strategy, excellent performance, and flexible deployment. It has found widespread applications in areas such as code assistants, intelligent customer service, and AI agent development.
• Industry analysts believe that Qianwen-3's explosive growth demonstrates that China's domestic large model open-source capabilities are catching up with international standards, further strengthening Alibaba's influence in the global AI foundational model ecosystem. 💪

Related Links:

• Tongyi Qianwen-3 Large Model Goes Viral Globally: Downloads Exceed 12.5 Million
• Alibaba Tongyi Qianwen-3 Model Breaks 12.5 Million Downloads in First Month of Open-Sourcing

英国金融监管局与Nvidia合作推出AI“超级沙盒”推动金融科技创新

简报：

• 英国金融行为监管局（FCA）宣布与Nvidia合作，推出“超级沙盒”计划，允许金融机构在监管保护下试验Nvidia的先进AI技术。
• 该平台将为获批公司提供实验环境，利用Nvidia高性能计算助力金融创新，如识别诈骗、股市操纵等风险防控场景。
• 项目预计于十月正式启动，旨在减少监管壁垒、促进商业和投资，并平衡监管和AI带来的岗位流失风险。
• 此举响应英国政府和工党推动AI开发与应用的相关政策目标，助力行业与消费者双赢并推动经济增长。

相关链接：

• aibase.com：英国金融监管局与Nvidia推AI超级沙盒

UK Financial Regulator Collaborates with Nvidia to Launch AI 'Super Sandbox' Driving Fintech Innovation

Briefing:

• The UK's Financial Conduct Authority (FCA) announced a collaboration with Nvidia to launch a 'Super Sandbox' program, allowing financial institutions to experiment with Nvidia's advanced AI technologies under regulatory protection. 🚀
• This platform will provide an experimental environment for approved companies, leveraging Nvidia's high-performance computing to foster financial innovation in risk control scenarios such as fraud detection and stock market manipulation.
• The project is expected to officially launch in October, aiming to reduce regulatory barriers, promote business and investment, and balance regulation with the potential risk of job displacement caused by AI. 💡
• This initiative responds to the policy objectives of the UK government and the Labour Party to promote AI development and application, fostering a win-win for both the industry and consumers while boosting economic growth. 📈

Related Links:

• aibase.com: UK Financial Regulator and Nvidia Launch AI Super Sandbox

英国多家机构致信监管机构反对Meta用AI自动化风险评估

简报：

• 多家英国互联网安全及儿童保护组织（如莫莉・罗斯基金会、NSPCC、Internet Watch Foundation）向英国通信监管机构Ofcom发信，呼吁限制Meta采用人工智能自动化平台风险评估流程。
• 这些团体担心，Meta公司计划让高达90%的风险评估工作由AI完成可能违反《在线安全法案》，影响对儿童保护和非法内容防控。
• 相关组织要求Ofcom明确定义，完全或主要自动化的风险评估通常不应被视为合格措施，并要求对平台简化风险评估的尝试提出质疑。
• Meta回应称，外界曲解其安全立场，公司仍在人类监督下使用AI工具帮助评估产品风险，确保合规。
• Ofcom表示正在考虑活动人士提出的担忧。

相关链接：

• 英国组织呼吁限制Meta用AI进行风险评估

UK Organizations Oppose Meta's AI-Automated Risk Assessment in Letter to Regulator 🛡️

Brief:

• Multiple UK internet safety and child protection organizations (such as the Molly Rose Foundation, NSPCC, Internet Watch Foundation) have sent a letter to the UK communications regulator Ofcom, calling for restrictions on Meta's use of AI to automate platform risk assessment processes. 🤖
• These groups are concerned that Meta's plan to have up to 90% of risk assessment work completed by AI might violate the Online Safety Act, impacting child protection and the prevention of illegal content. ⚖️
• The organizations demand that Ofcom clearly define that fully or predominantly automated risk assessments should generally not be considered qualified measures, and urge scrutiny of platforms' attempts to simplify risk assessments.
• Meta responded, stating that external parties have misrepresented its safety stance, and the company continues to use AI tools under human oversight to help assess product risks and ensure compliance.
• Ofcom stated it is considering the concerns raised by the activists.

Related Links:

• UK Organizations Call for Restrictions on Meta Using AI for Risk Assessment

谷歌Gemini应用月下载量首超ChatGPT，用户日活率仅为4.9%

简报：

• 截至2025年4月底，谷歌Gemini应用在全球Android设备上的月下载量首次超过ChatGPT，每周安装量超过600万，而ChatGPT同期下载量降至300万以下；
• 此增长被认为与谷歌通过其搜索引擎和其他服务的推广有关；
• 尽管下载量显著提升，2025年5月数据显示Gemini每日活跃用户比例仅为4.9%，远低于ChatGPT的42.52%，反映出Gemini在用户粘性和日常活跃度上与主要竞争对手仍有较大差距。

相关链接：

• https://www.aibase.com/zh/news/18737

Google Gemini App's Monthly Downloads Surpass ChatGPT for the First Time, Yet Daily Active User Rate Stands at Just 4.9% 📉

Brief:

• As of the end of April 2025, Google Gemini app's monthly downloads on Android devices globally surpassed ChatGPT for the first time, with weekly installations exceeding 6 million, while ChatGPT's downloads in the same period dropped below 3 million. 📈
• This growth is attributed to Google's promotion through its search engine and other services;
• Despite the significant increase in downloads, May 2025 data shows Gemini's daily active user (DAU) percentage is only 4.9%, significantly lower than ChatGPT's 42.52%, indicating a substantial gap between Gemini and its main competitor in terms of user stickiness and daily activity. 😮

Related Links:

• https://www.aibase.com/zh/news/18737