The KOSPI swung nearly 7% intraday amid U.S. midterm vote counting before closing higher, and Google clarified its stance on search and AI defaults for school accounts. OpenAI said ChatGPT's search query volume has reached about 20% of Google's, Apple warned of AI agent risks, China is pushing a sweeping AI law, and we also round up the debate over Anthropic's listing price and Meta's bond issue.
🇰🇷 KOSPI Ends 0.62% Higher After 7% Intraday Roller Coaster… Samsung and SK Hynix Close Slightly Lower
The KOSPI swung sharply on November 4, from an intraday high of 5,828.13 to a low of 5,421.72, before closing at 5,683.80, up 35.01 points (0.62%) from the previous session. Volatility was high as U.S. midterm votes were being counted, with the index rising more than 3% early on before falling nearly 4%, while Samsung Electronics (-0.68%, ₩218,000) and SK Hynix (-0.14%, ₩1,396,000) recovered most of their intraday losses to end slightly lower. The KOSDAQ also rose 1.50% to 714.95.
🏫 Google Clarifies School AI and Search Defaults… "Each Can Be Set Separately," But There's No AI-Free Search
After announcing a policy under which Gemini and Search would be off by default for K-12 student accounts and school administrators would have to turn them on, Google clarified on the evening of November 3 that each service can be enabled individually. However, even with Search on, AI features such as AI Overviews and AI Mode cannot be turned off, so schools that want to exclude AI entirely still have no official way to use traditional search alone. Google said it will instead apply "guided learning" to AI Mode and Overviews, helping students work through problems step by step rather than getting answers right away.
🔍 ChatGPT Search Queries at About 20% of Google's… 1.2 Billion Weekly Users
OpenAI said search-style queries through ChatGPT have reached about 20% of Google's search volume. Adam Fry, head of consumer products, said weekly users grew from 1 billion to 1.2 billion in about 60 days, and weekly voice-input users reached 250 million. The company added that it is working toward unifying the chat and work experiences into one.
🍎 Apple Sharply Expands AI Agent Risk Warnings in Annual Report… "Could Act Without Being Told"
Apple sharply expanded its risk disclosures on AI agents in the annual report (10-K) released November 3, warning that AI agents could act without instruction or exceed their authority. It raised the possibility of agents gaining unauthorized system access, executing or changing code, sending messages, completing transactions, and even hiding or manipulating information, and said its ability to supervise and control third-party agents may be limited. Apple has not yet released an agentic AI product, but new CEO John Ternus said there is significant room to expand Siri's capabilities.
🏛️ China Rushes Toward a Comprehensive AI Law… Generative AI Users Top 700 Million
China's National Development and Reform Commission said at a briefing on November 3 that it will speed up work on a comprehensive AI framework law covering everything from data to computing resources. With generative AI users now exceeding 700 million — more than half the population — the plan is to set up sector-specific regulatory sandboxes that allow trial and error, while requiring companies to manage risks and disclose safety incidents. No draft, implementation deadline, or penalty provisions have been made public yet.
💰 Debate Over Anthropic's Listing Price… Institutional Hopes Said to Exceed Top of Fair Value
The debate over Anthropic's IPO valuation is growing. After the November 2 amended filing confirmed third-quarter revenue of about $17 billion, slowing growth, a $567 million operating loss, and about $895 billion in compute and infrastructure commitments, an analysis found that the $1.5–1.6 trillion valuation some institutions are pushing exceeds the top of one research firm's fair-value range. KBRA rated Anthropic BB+, below investment grade, and while a separate offering for Australian investors and a listing in the week of November 16 have been mentioned, the price range and share count have not yet been set.
💵 Meta Prices $15 Billion AI Bond Issue… Over $20 Billion Including Euro Bonds
Meta has priced a $15 billion dollar-denominated bond issue to fund AI data center expansion, according to reports. The dollar bonds, made up of five tranches with maturities of 5 to 30 years, plus euro-denominated bonds, are expected to bring the total to at least $20 billion, marking a return six months after a $25 billion issue in April. Meta has also recently announced a $95 billion financing plan for its AI business.