已验证来源
每个故事都链接到最有力的可用证据：可用的原始来源，否则明确归因的报告。
简单的英语
发生了什么、为什么重要以及值得关注的内容——无需行话。
无填料
当信号很弱时，我们不会发布任何内容，而是填充提要。
更多故事
9 故事安全
Hugging Face Chat interface vulnerability bypassed admin API via URL encoding
A security flaw in the open-source Hugging Face Chat-ui allowed unauthorized access to administrative functions by using percent-encoded characters to bypass path-based security filters.pasqualepillitteri.it工业
KOSPI index drops 6% amid investor concerns over U.S. AI data center spending
South Korea's KOSPI index fell 6.02% on November 6, 2026, as investors reacted to potential shifts in U.S. AI infrastructure policy following midterm election results.en.yna.co.kr工业
NUHS, Singtel, and Ericsson launch research centre for AI-driven 6G healthcare
A new research centre in Singapore is testing AI applications for home-based patient care and robotic hospital logistics, leveraging 5G-Advanced and future 6G network capabilities.straitstimes.com创新
Sunderland hub receives £12.2m to use AI in eye scans for disease detection
The Northern Ophthalmic Research and Innovation hub at Sunderland University has secured government funding to develop AI tools that analyze routine eye scans to detect early signs of diabetes and heart disease.bbc.com政策
Bombay High Court quashes customs order over alleged AI hallucinations
The Bombay High Court set aside a customs adjudication order after a logistics company alleged the Commissioner relied on non-existent Supreme Court judgments generated by AI, citing a recent Supreme Court precedent demanding zero tolerance for unverified AI-generated precedents.jurishour.in创新
Cross‑agent LLM verification shows limited agreement even with top‑tier models
A new arXiv paper introduces a protocol where one LLM translates clinical‑trial criteria into a Dafny predicate and another LLM implements a method checked against it, revealing that agreement rates stay well below a same‑model baseline.arxiv.org工业
IIT Delhi launches six-month certificate program in AI security
IIT Delhi has introduced a new six-month certificate program focused on cybersecurity and AI security, targeting working professionals with a curriculum covering adversarial machine learning and agentic AI exploitation.educationtimes.com政策
Analysts say US midterm ‘blue wave’ will slow but not stop AI data‑center buildout
Nikkei Asia reports that analysts expect the Democratic gains in the recent U.S. midterm elections to temper the pace of AI data‑center construction, offering a modest boost to Asian tech firms seeking to expand their supply chains.asia.nikkei.com企业
Michigan Medicine names first chief data and AI operations officer
Michigan Medicine has appointed Naveen Maram as its inaugural chief data and artificial intelligence officer, a role aimed at steering AI integration across patient care, operations, and research.michigandaily.com