Brockman 对谈 a16z：AGI 已发生，就业与安全无人有答案，AI 真正形态未现。

Binance Blog published a new article, revealing how its AI-driven risk systems are being used to protect users, reduce fraud, and strengthen compliance and internal operations. In the first half of 2026, Binance said these systems helped protect more than 8 million users and prevent approximately $4.6 billion in potential losses. The company said the protections covered abnormal trading activity, account takeover attempts, scams, and transaction fraud. Binance also said AI now informs 80% to 90% of real-time risk decisions across identity verification, account security, payments, and transaction screening, while human reviewers handle edge cases that require more detailed judgment. The article explains how in-house and external models work together, why human oversight remains important, and how the same approach is being applied beyond loss prevention to compliance and internal workflows. Binance said it intercepted millions of scam and phishing attempts in H1 2026, blacklisted more than 42,000 malicious addresses, and issued over 14,000 real-time warnings daily. The company said these results are supported by more than 100 AI models used across anti-fraud and anti-scam controls. Human risk analysts set thresholds, review unusual cases, and retrain models as new scam patterns appear. In KYC identity verification, Binance said AI-enabled review pipelines have delivered up to 100x operational efficiency gains over manual processes in specific workflows, while specialists remain involved in higher-risk cases. The article also said AI is embedded across the user journey, including identity verification, account security, payments, and broader transaction protection and screening, allowing most checks to be resolved automatically without slowing legitimate users. Binance described its approach as a hybrid model stack that combines proprietary systems with external AI and foundation models, while an internal Red Team tests defenses against realistic attack methods. Chief Security Officer Jimmy Su said these exercises help identify weaknesses before attackers do and strengthen the people, processes, and technology protecting users.

黄仁勋与 Dario 互怼都靠归纳和外推，作者借休谟拆解 AI 末日论之争：感恩节还没到。

According to Wallstreetcn, Canada’s stock market edged lower at the Toronto open, with the S&P/TSX Composite Index down 0.4% at 9:53 a.m. Toronto time, while energy was the only sector to rise, supported by higher oil prices.

KPMG will cut about 200 jobs in its UK consulting business, including roles related to artificial intelligence. According to Sina Finance, the planned cuts are concentrated in the data and technology division and account for about 4% of KPMG's total UK consulting workforce. People familiar with the matter said the affected employees are expected to leave in October after the current consultation process is completed. The sources said the company told staff it needs to carry out "scale optimization" in the affected business units. About six months earlier, KPMG UK had already faced the risk of cutting nearly 600 jobs, including more than 400 in audit and about 120 in consulting. At the time, weak natural attrition and soft market demand forced the firm to reduce headcount. Like its three other Big Four peers, KPMG's consulting business has also faced a prolonged slowdown in demand.

Agility Robotics Inc. on Tuesday unveiled a new humanoid designed to work safely in close proximity to people, according to Bloomberg. The launch marks a milestone for an industry that typically keeps robots separated from human workers.

金色财经推出《金色Web3.0日报》，为您提供最新、最快的AI、DeFi、DAO、NFT和元宇宙行业新闻。

Anthropic、OpenAI等罕见同频呼吁放缓AI，却同时冲刺史上最大IPO。

黄仁勋在 All-In Summit 点名炮轰 AI 末日论：这些预测全都错了；演讲中途特朗普

多篇研究聚焦AI递归自我改进风险，越来越多前沿研究者担忧失去控制，学界分歧加剧

Robinhood 链上股票 DEX 63 天成交破 30 亿美元，AMC 董事长开炮质疑无需公司许可。

进入2026年，A股已成妖股天下。一只股票一年暴涨逾28倍，幕后谁在爆炒？击鼓传花…

从 800 万美元的恐惧广告到上亿浏览的派系骂战：资本在幕后采购叙事、分发焦虑

Dario Amodei 首认行业长期隐瞒 AI 风险，提出三步走限速框架，遭开源阵营激烈反弹。