Pre-market US stocks in the storage, optical communications, semiconductor, and Neocloud sectors mostly rose, with SNDK up 2.74%.

Beating AI Insight News: OpenAI announced the appointment of former U.S. Army Secretary Dan Driscoll as its new Senior Advisor for National Security Modernization. Driscoll will assist the company in collaborating with the U.S. government and its allies, drive AI applications in defense, cybersecurity and other fields, and help OpenAI understand the military’s actual operational needs. A veteran of the Iraq War, Driscoll resigned earlier this year due to disagreements with Defense Secretary Pete Hegseth. During his tenure, he promoted an initiative called Operation Jailbreak, which aimed to connect the long-incompatible radar, weapons and battlefield command systems of the U.S. military to enable cross-system data sharing. OpenAI previously explicitly banned military uses in its usage policies, but removed this blanket restriction in early 2024. In 2025, the company secured a U.S. Department of Defense contract worth up to $200 million. This February, OpenAI announced it would integrate ChatGPT into the GenAI.mil platform, which serves around 3 million military and civilian personnel, and signed an agreement with the Pentagon to allow its models to be deployed in classified environments. The military collaboration has sparked internal controversy: in March, OpenAI’s head of robotics and hardware Caitlin Kalinowski resigned over concerns that the Pentagon agreement was advancing too hastily and lacked clear security safeguards. OpenAI stated that the agreement prohibits using its technology for large-scale domestic surveillance or allowing AI to independently control autonomous weapons.

According to an announcement on the ZK Nation forum, Matter Labs has completed the SOC 2 Type II audit of its Prividium platform, covering four areas: security, availability, confidentiality, and privacy. The audit was conducted by Thoropass Assurance, an independent audit firm recognized by the American Institute of Certified Public Accountants (AICPA). Earlier this year, a SOC 2 Type I audit was completed, which verified the reasonable design of the platform’s control measures as of May 29. The Type II audit further validates whether these control measures have operated effectively over a continuous period, covering the timeframe from July 1, 2026 to September 30, 2026. The announcement notes that for banks and regulated financial institutions, SOC 2 Type II reports are typically required by vendor risk assessment teams before a platform is deployed to production environments. Currently, the report is accessible to existing and qualified prospective clients after signing a non-disclosure agreement (NDA). This SOC 2 certification is also one of the deliverables outlined in the TPP-18 proposal. Matter Labs previously stated in its Q3 2026 delivery report that the Type II audit was still in progress, and the work has now been formally completed.

According to official sources, Bitget’s US Stock IPO Prime platform is set to launch an Anthropic IPO subscription channel, enabling global users to access investment opportunities spanning the pre-IPO stage through the initial public offering. This offering opens institutional-grade IPO allocation quotas, supporting subscriptions via USDT or USDGO; allocated assets will be delivered as rANT, which leverages Bitget’s RWA (Real World Assets) infrastructure to mirror the post-listing economic performance of ANT on a 1:1 basis. As announced, Anthropic formally filed its US IPO prospectus on October 26, 2026, with plans to list on the Nasdaq under the ticker symbol ANT. Current market projections value the IPO at approximately $1.5 trillion to $1.75 trillion. The subscription is divided into two rounds, with details as follows: ? VIP Round: Open to VIP1-VIP7 users. Subscription period: November 5, 16:00 to November 9, 16:00 (UTC+8). Investment range: $100–$3,000,000. Total investment pool cap: $1,000,000,000. Historical reference allocation rate: 40–80%. Underwriting service fee: 12%. Distribution time: 3–5 business days after the IPO. ? Public Round: Open to all eligible users. Subscription period: November 10, 16:00 to 16:00 on the day prior to the IPO (UTC+8). Investment range: $100–$25,000. Total investment pool cap: $200,000,000. Underwriting service fee: 5%. Distribution time: 2–3 hours before the opening of regular US stock trading on the IPO day.

Blockstream co-founder Adam Back published an article discussing the group supporting the BIP110 fork, and mentioned Francis Pouliot’s discussion on the real-world state of Bitcoin, noting that it once again clearly demonstrates the actual operational logic of the Bitcoin network. BIP110 suffered a severe failure during its initial technical review, with the technical proposal containing numerous flaws and being highly problematic. However, the proposal’s supporters did not opt to listen to feedback and revise the plan, instead attempting to push it through by force. In response, Adam Back stated that any attempt to impose personal views on Bitcoin will ultimately be rejected by the market. He also noted that the related memes about "Big Bitcoin" (BIG BITCOIN), while well-intentioned, aim to help relevant groups understand Bitcoin’s immutability.

Beating AI News Flash from Dong察: Deep Principle has released its research model Hydrogen-1.0, trained on Zhipu AI’s GLM-5.3. In OpenAI’s FrontierScience-Research benchmark test, Hydrogen-1.0 scored 38.3 points, outperforming GPT-6 Astra’s 33.3 points, while the original GLM-5.3 only scored 24.4 points. Its training data is unique: Scientific papers typically only present final results, but real research also requires understanding how experiments are designed, where errors occur, and how to adjust next steps. Deep Principle used its own lab’s complete experimental records—including the full process of trials and revisions—to train the model. In practical demonstrations, Hydrogen-1.0 can write its own code and call scientific software to simulate the movement of plastic molecules. If it detects errors in simulation settings midway, it will self-repair and resume running and analyzing results. In another demo, it developed experimental protocols for 12 chemical reactions, and noticed a missing pipette tip replacement step in the virtual lab workflow, immediately revising the process. The team also had it conduct five consecutive rounds of new material screening. Each round determined the next set of trials based on the evaluation results of the previous round. When searching for metal complexes with target electronic properties, the candidate material it ultimately found exceeded both GPT-6 Astra and Kimi-K3 in the metrics set by the team. The model’s weights are open-sourced under the MIT License, but the lab data used for training has not been made public.

Santiment stated in a post that since the Binance hack, the crypto exchange’s ETH supply has declined in two phases, only one of which aligns with the timing of the hacker’s fund transfers. Data shows that as of the close of October 28, ETH held by exchanges made up 3.56% of the total ETH supply; by the close of November 2, this share had dropped to 3.20%. Over half of this decline occurred on October 29, the day Binance was hacked. TRM Labs data indicates that roughly 200,630 ETH was withdrawn from Binance’s hot wallet on October 29. However, ETH holdings of major exchange addresses did not fall that day—instead, they increased by around 39,000 ETH. Then, from November 1 to November 2, as exchanges resumed withdrawals, these addresses saw their ETH holdings decrease by approximately 313,000 ETH. From the close of October 28 to November 2, major exchange addresses’ total ETH holdings fell by a cumulative 297,000 ETH, a drop of about 5.4%. Santiment added that based solely on exchange data, it is currently impossible to determine where the ETH ultimately ended up, nor whether it was moved to cold wallets or transferred to other trading platforms.