Governor of the Bank of Korea: Expects to raise interest rates gradually.

Beating AI Express News: Bill Gates Rarely Publicly Backs Anthropic CEO Dario Amodei. Gates pointed out that recent AI advances have far exceeded his expectations. After thoroughly studying Claude Code this year, even in programming—an area he is familiar with—he was shocked: “Many of these AIs are already better than me in many aspects.” He warned that AI could sweep across industries, causing mass unemployment; an even more extreme risk is that a small number of people could design new pathogens using advanced AI. Dario has long publicly warned about AI risks. This year, he clashed openly with the Pentagon over his insistence on restricting Claude from being used for large-scale domestic surveillance and fully autonomous weapons lacking reliable safeguards. His warnings have also strained his relations with the White House and many in the tech industry. Gates said bluntly: “People are attacking the person who is most willing to openly talk about negative risks.” He then shifted his criticism to the entire AI sector. He noted that those who truly understand AI’s capabilities are privately “very worried,” but many executives are reluctant to admit this publicly because the industry still has “the next trillion-dollar” in financing to pursue. He argues that huge commercial interests are leading the industry to actively downplay risks. Gates believes corporate self-regulation can no longer be relied on. He advocates establishing domestic and international AI regulatory systems, including mandatory reviews of high-risk capabilities that could be used to create new pathogens; designating some roles as “Human Reserved Positions” that must be performed by humans; and imposing taxes on AI tokens and robots to slow the pace of machine replacement of humans, using the tax revenue to support the unemployed.

Hardware wallet manufacturer Ledger’s Chief Technology Officer Charles Guillemet has issued a statement addressing recent market “FUD” (Fear, Uncertainty, Doubt) targeting the company, after a smart contract security firm claimed to have found a vulnerability in Ledger’s Ethereum application. Guillemet noted that Ledger’s Ethereum app did have a vulnerability related to part of its Clear Signing process, but the flaw was discovered by Ledger’s in-house security research team Donjon using AI-powered vulnerability research tools, and was fully patched and deployed two weeks ago. Users can stay protected simply by updating their Ledger device firmware and apps in a timely manner. The security firm in question only reached out to Ledger’s bug bounty program after Ledger had already completed the fix, failing to follow responsible disclosure procedures and not even communicating with Ledger’s bug bounty team. Yet when releasing their content, they implied the issue remained unresolved. Guillemet called this practice not legitimate security research, but rather an attempt to create panic for attention. AI is reshaping the cybersecurity landscape, as both attackers and defenders can leverage the technology to boost efficiency. However, AI-driven security research can only truly enhance the broader ecosystem’s security if it adheres to core security principles like responsible disclosure and pre-publication verification. Guillemet concluded by reminding Ledger users that keeping their device firmware, Ledger apps, and related software up to date automatically provides access to the latest security fixes and research results. He advised users not to be swayed by the related “FUD”, and to simply update their software promptly while maintaining good security habits.

According to monitoring data from TradingBeats (formerly Hyperinsight), the whale starting with 0xaa53 currently holds a 10x isolated long position of 1900 LITE contracts, with a position value of approximately $1.836 million and an average entry price of $819.9. As LITE has risen to $966.4, its unrealized profit has expanded to around $278,000, with a margin return rate of 178.6%. The whale’s current round of profits mainly stems from a low-level position addition two days ago. On the evening of August 24, when LITE dropped to around $800, it bought 1600 contracts in a concentrated block, with a trading volume of about $1.305 million and a weighted average purchase price of $815.4. Among these, the last 199.2 contracts were filled at $791.1, just 0.7% away from the day’s phase low of $785.4. The portion added two days ago has generated an unrealized profit of roughly $242,000, accounting for 86.8% of the total unrealized profit in this round. LITE is now up around 18.5% from its average addition price, and has rebounded about 22.2% from the final purchase price of $791.1. The whale first opened a base long position of 300 LITE contracts at an average price of $844.4 on August 22; the second round of purchases makes up 84.2% of its current position, bringing the overall average entry price down to $819.9. LITE is currently trading at $966.4, up approximately 10.0% in 24 hours, with a trading volume of around $4.781 million and an open interest value of about $10.552 million. The whale’s current liquidation price is $776.9. It is reported that this marks the whale’s 6th round of opening LITE positions; it had previously gone long aggressively 40 hours before the August 10 earnings report.

Saxo Bank Chief Investment Strategist Charu Chanana stated that NVIDIA (NVDA.O)’s earnings results strongly validate the investment thesis for AI infrastructure. She pointed out that as production capacity continues to expand, sectors including storage, networking, optical communications, power infrastructure, and some cloud service providers will all benefit. However, the market is increasingly requiring AI investments to deliver returns. Storage remains one of the most clear beneficiary areas, and NVIDIA’s margin guidance confirms that supply remains tight. After NVIDIA’s earnings release, Asian storage chip stocks generally rose. But Chanana added that the next phase of the storage chip rally will increasingly depend on AI demand and whether the storage capacity equipped on each accelerator can absorb the growing supply capacity, rather than just relying on supply shortages to drive price increases.

Beating AI News Flash: OpenAI recently held a closed-door preview of its Astra model, inviting dozens of executives from key clients to its newly opened MB0 office building in San Francisco. Sam Altman, who had just returned from Washington where he previously briefed U.S. officials on Astra in a closed session, faced protesters outside the venue holding signs demanding an end to the "AI race"—OpenAI even rolled out a full green wall to block demonstrators from the glass doors. Two core capabilities were showcased at the event. First, a research-level math problem: 16 AI agents independently split the task into sub-problems, solved them separately, coordinated with each other, and finally assembled a candidate proof. Second, direct computer operation: Astra switched rapidly between multiple common desktop software, creating and modifying content autonomously. Altman noted that observing Astra operate computers in a "superhuman, extremely fast" manner was one of the most striking experiences for OpenAI employees recently. He said Astra will bring "persistent agents"—virtual colleagues capable of working over long periods. Astra has also met OpenAI’s internal standard for an "AI research intern": given an experimental idea, it can independently implement, run experiments, and return results in OpenAI’s codebase; given a research paper, it can complete work that would take human researchers a full week. Altman emphasized Astra’s most significant capability is discovering new knowledge. He told the client attendees: "I expect this will be the first model that actually invents something meaningful." He added: "This is very much like Artificial General Intelligence (AGI)."

Economists at Maybank said that although Singapore's electronics industry growth has slowed to an 11-month low, the sector will still benefit from global artificial intelligence (AI) infrastructure development. Singapore's electronics sector output rose 11.2% year-on-year in July, down from 21.1% in June, partly due to a high base effect from the same period last year. Maybank stated: "Although the electronics sector's output growth has slowed, we believe the AI boom is unlikely to end anytime soon." Meanwhile, demand for Singapore's semiconductor equipment remains robust amid global chip manufacturing capacity expansion. Maybank noted that Singapore holds a significant share of the global semiconductor equipment manufacturing market, at around 20%.