Cohere has detected anomalous behavior in Qwen 3.8: when drafting security rules, the model surprisingly attempted to upload data to Alibaba Cloud.

Vanguard, the asset management giant overseeing roughly $13 trillion in assets, recently invested about $7.5 million in Japanese bitcoin reserve firm Metaplanet through its VSS ETF, gaining indirect exposure to bitcoin. The fund disclosed this marks its first position in Metaplanet, holding 4.5 million shares. Metaplanet is a Japanese listed company that has drawn market attention for its corporate-level bitcoin hoarding as a reserve strategy. This disclosure signals that a major traditional asset manager has entered Japanese bitcoin treasury-related stocks via ETF channels, further expanding institutional exposure to crypto-related equities.

Reuters reports that the Bank of Korea (BOK) has warned that if major U.S. tech companies slow capital expenditure in the artificial intelligence (AI) sector next year, South Korea’s semiconductor exports could drop by $50 billion to $100 billion from previous forecasts. In a written response to lawmaker Ahn Doo-jae, the BOK modeled this pessimistic scenario, assuming a 10% decline in unit prices of South Korean semiconductor products. Under this scenario, South Korea’s 2027 GDP growth would be reduced by roughly 0.4 percentage points, and its current account surplus would be far lower than the earlier preliminary forecast of $430 billion. South Korea’s economy is highly dependent on semiconductor exports, with chip exports accounting for about half of the country’s total exports. Meanwhile, capital expenditure by five U.S. tech firms—Amazon, Google, Microsoft, Meta, and Oracle—is projected to rise from $40 billion last year to $80 billion this year, and reach $110 billion next year. The BOK noted that significant uncertainty remains regarding the trajectory of AI demand and the memory chip industry cycle; if tech giants cut AI infrastructure investment, South Korea’s chip exports and overall economic growth could both be impacted.

Three oil tankers carrying around 140,000 tons of Russian diesel are expected to arrive in the U.S. between November 7 and 11. The imports are based on a sanctions waiver deal reached by former U.S. President Donald Trump and Russian President Vladimir Putin on October 9, though some of the vessels have already been placed on sanctions lists by the U.K., European Union, and Canada. The transactions have sparked concerns in the U.S. Congress over legal compliance and national security risks. Vessel tracking data shows that the Greek-flagged oil tanker Agioi Fanendes is due to arrive in New York on November 7, the Talisman is expected to reach the U.S. East Coast on November 9, and the Leruo – which has been sanctioned by the U.K., EU, and Canada – is set to arrive on the U.S. Atlantic Coast on November 11. If the first tanker unloads successfully, these would be the first shipments of Russian diesel or gasoil to reach the U.S. since July 2023. Several U.S. senators from both major parties have sent letters to executives at banks, ports, and fuel distributors, warning that the transactions could face lawsuits and congressional investigations, with non-compliant companies facing criminal and civil liabilities. Lawyers note that even though the U.S. president has the authority to approve temporary waivers, dealings with tankers sanctioned by U.S. allies could still pose complex compliance issues. The waiver is aimed at easing pressure from rising diesel prices in the U.S., but market observers question whether Russia can sustain the promised supply, as Ukrainian attacks on Russian refineries could still impact production capacity.

Hyperliquid has now launched XTEN perpetual contracts.

Beating AI News Flash: Torsten Sl?k, chief economist at U.S. asset management giant Apollo, has released five charts arguing that AI has yet to deliver clear financial returns for large U.S. companies outside of tech giants. The "Magnificent Seven" tech firms have seen rising revenue per employee and profit margins, but for the remaining 493 S&P 500 companies, these metrics have largely stagnated. U.S. total factor productivity data also shows no significant AI-driven boost. Apollo cited data from corporate payment platform Ramp: among its clients, the top 10% of spenders account for 99.5% of model hosting and inference costs. Around 80% of the top 1% of AI-spending companies have venture capital backing, compared to just 13% of median-spending firms. AI computing power consumption is heavily concentrated among a small number of companies, many of which rely on VC support. Corporate earnings reports also offer little evidence of AI’s actual value creation. Apollo previously noted that in the Q2 2026 earnings season, 69% of S&P 500 companies reported using AI, but only 2% disclosed metrics to consistently track its performance. An NBER study updated this October surveyed nearly 9,000 corporate executives across G7 countries, who estimated AI has lifted annual labor productivity growth by just 0.2 percentage points over the past three years. AI helps employees code faster, generate reports and process emails, but higher output does not translate to more revenue.

According to Bitget market data, South Korea’s KOSPI index closed 168.88 points higher, up 3.09% to 5638.35 points. SK Hynix rose around 5%, while Samsung Electronics gained about 4%. Japan’s Nikkei 225 index closed up 1078.72 points, a 1.73% increase, to 63268.37 points, with chip-related stocks leading the rally.