US stocks were broadly lower in pre-market trading, with optical communications stocks, storage stocks, and the Neocloud concept all declining, none spared.

South Korean crypto exchange Bithumb has issued an announcement updating the aftermath of its recent incident. On October 27, unauthorized external personnel breached the company’s internal mobile office system (Daou Office), leading to the leakage of some personal information. The system is fully separated from exchange accounts and asset data, and no direct breach of the trading system or customer database has been detected so far. Bithumb has since strengthened real-time monitoring of abnormal transactions and withdrawals. The exchange disclosed that the total number of affected individuals includes 6,444 people: 5,600 of whose leaked data is encrypted and unreadable, while 844 have readable information exposed. Additionally, 715 employees and partner personnel had their information compromised. Stolen data includes names, phone numbers, emails, resident registration numbers, driver’s license numbers, bank accounts, and ID card copies. Bithumb emphasized that it immediately blocked access to the relevant system, deactivated it, and filed incident reports with the Korea Internet & Security Agency and the Personal Information Protection Commission. The company detected the unauthorized access to its internal mobile office collaboration system on October 27, and is currently analyzing the full scope of the leak, including whether member information is involved.

Fomo, an early investor and advisor (hdf, @PaikCapital), has joined the trading app Fomo as Head of Perpetual Contracts. He noted that with a decade of trading experience, his goal in joining Fomo is not merely to make perpetual contract trading accessible to all, but to build a network that elevates the proficiency of every trader.

Beating AI Express News: Independent developer Divyansh Agrawal has built Qwenputer based on Alibaba’s Qwen3.5-4B, integrating a 32-bit RISC-V virtual CPU directly into the model’s weights. This allows the large language model (LLM) to perform precise calculations when generating responses without relying on external computing tools. Instead of retraining Qwen, Agrawal added 11 layers before the original 32-layer network, using a compiler to translate the CPU’s operation rules directly into model parameters. The added network leverages the Gated DeltaNet architecture to implement precisely readable and writable memory for storing programs and data. The virtual CPU can now run compiled C programs and has passed RISC-V instruction tests, with no errors detected in tests executing approximately 18 million total instructions. In DeepMind’s 40 large-number multiplication problems, the original Qwen got 22 correct answers, while Qwenputer scored 39; the original Qwen achieved a perfect score when connected to external Python tools. The CPU itself executes instructions accurately, but the LLM may still make errors when calling it or outputting results. This technical approach has precedents: in March this year, Percepta implemented a virtual computer capable of running C programs using a neural network, though it was a purpose-built model. By contrast, Qwenputer integrates similar computing capabilities into the already trained Qwen, enabling the LLM to use its internal CPU for calculations during response generation. However, the solution is currently very inefficient: a large number of zero-value parameters expand the model file to 18.8GB, roughly twice the size of the original, and its computing speed is far slower than external tools.

According to relevant pages, the Binance Wallet PREDICT token presale has raised 197,300 BNB, with a target of 7,794.12 BNB, currently oversubscribed by approximately 25 times. Earlier reports noted that the participation threshold for predict.fun’s (PREDICT) Pre-TGE Prime Sale was 60 points, and 15 points will be deducted for joining the current TGE event. The event runs from 16:00 to 18:00 today, with a per-user subscription limit of 6 BNB.

After yesterday’s FOMC meeting, Tom Lee told CNBC that Federal Reserve Governor Christopher Waller’s latest remarks leaned dovish, which bodes well for risk assets and a rebound in AI-related trading. Lee pointed out that Waller publicly challenged the claim that “AI growth will inevitably lead to inflation” — a critical point, as markets had previously fretted that AI capital expenditures would drive up inflation, forcing the Fed to raise interest rates or tighten monetary policy. “Therefore, now we can treat good news as good news. You will be able to say that significantly better-than-expected data is a positive, because the Fed will not attempt to slow artificial intelligence’s growth,” Lee added.

According to Reuters citing sources familiar with the matter, London Metal Exchange (LME), a subsidiary of Hong Kong Exchanges and Clearing (HKEX), is exploring tokenizing metal inventories stored in Hong Kong warehouses to turn them into digital assets eligible for loan collateral and other financing activities, in line with Hong Kong’s strategy to develop into an international commodity trading hub. LME responded that it is currently in the early stages of exploring tokenizing metal inventories for which warehouse receipts have not been issued. Unlike metals under warehouse receipts that already have standardized financing mechanisms, these inventories currently lack mature financing channels. Sources revealed that LME plans to hold a closed-door industry roundtable in early November to discuss the plan. HKEX is also testing carbon credit tokenization to enable near-real-time transaction settlement.