RSS feed for Geopolitics & Trade →
Ting-Wei Sun’s guilty plea puts a concrete legal marker in a case that has made the AI-server supply chain look less like a simple sale and more like a sequence of decisions that must withstand scrutiny.
A prompt that appears to be headed for Claude may pass through another service first, and that route can determine whether the person who sent it knows who can read or retain the exchange. In its threat-intelligence report, Anthropic alleges that some paths into its model used fake accounts, stolen credentials, and proxy services.
Synopsys is looking into exploring partnerships with Chinese AI labs that could help Chinese companies design chips faster with AI-assisted versions of its tools.
CrowdStrike has supplied the missing technical layer in South Korea’s bank-intrusion inquiry: a trail of AI-tool artifacts that it says appeared in infrastructure controlled by an unidentified attacker. The disclosure moves the public discussion beyond a general warning that AI may have figured in the incidents.
CCA argues that a group of Chinese AI developers is responding to constrained access to advanced processors by redesigning how models consume compute. Sparse attention, mixture-of-experts systems, and memory-saving approaches can make a fixed pool of hardware do more work, making raw GPU counts an incomplete guide to model-serving capacity and deployment.
Butterfly Effect has completed a financing round of more than US$500 million for Manus, giving the AI-agent startup new backing after Meta unwound its prior acquisition.
StepFun’s Alliance Makes the Compatibility Question Concrete For readers trying to judge how export controls affect Chinese AI development, the useful starting point is not a claim that a domestic chip has solved every problem.
A US$92,478 investment in a Shanghai robotics company has resulted in a US$200,000 civil penalty for Amidi, LLC, a sharp practical warning for investors whose deal activity runs through foreign-controlled vehicles. On October 7, the Treasury Department announced the first civil penalty under its Outbound Investment Security Program, or OISP.
South Korea’s Financial Services Commission has made a bank intrusion reported by Shinhan Bank the starting point for a broader test of how the country’s financial system responds when artificial intelligence may be part of an attack.
President Donald Trump formally announced the Super Intelligence Force on October 4, giving a name and leadership roster to a federal task force proposed weeks earlier. ABC News reported that Jay Clayton, Andrew Ferguson, Emil Michael, and Scott Kupor were named to lead it.
Reflection AI just introduced Beam as its first open-weight model, putting it forward for coding, reasoning, and agentic workloads. The announcement did more than outline a new model: it named Z.ai’s GLM-5.2, Alibaba’s Qwen 3.8-Max, and Kimi K3 as the points of comparison. Reflection said Beam remained in final red-teaming and evaluation.
Huawei and Qualcomm announced a multi-year broad patent-license agreement on Oct. 5 that cross-licenses 5G, compute, artificial intelligence, and networking. In the same announcement, Qualcomm said it intends to purchase certain Huawei U.S. patents in compute, AI, networking, and other technologies.
The Horizon3 disclosure of a vulnerability in Rejetto HFS was followed within days by exploitation attempts against vulnerable hosts, including activity sourced from a China-hosted IP.
The latest US-China RISC-V story is not a bilateral chip deal. It is a standards and ecosystem story, where engineers and organizations can participate in the same open architecture even while Washington and Beijing remain strategic rivals.
China’s access to immersion deep-ultraviolet lithography is at the center of an export-control debate after a South China Morning Post account highlighted a Center for Technology & Statecraft estimate of 343 such systems at Chinese-owned fabs by early 2026. That number is consequential precisely because it is not a confirmed census of operating equipment.
Filings reviewed by Bloomberg describe a financing trail in which Semi-Tech Leasing Group Co. supplied capital to Glory View Technology Co. for purchases of more than 700 servers over roughly the preceding year. According to Bloomberg’s account, the documents point to more than 3 billion yuan in borrowing through May 2026.
Federal prosecutors have outlined an alleged export-control scheme centered on a California technology company, high-end AI servers, and routes through Malaysia and Singapore. On October 1, the U.S. Department of Justice said it had arrested 38-year-old San Gabriel resident Greg Lui, whose court identification also includes Yiu Kong Lui, under a three-count federal indictment.
The United States has put a narrow but consequential idea on the table in its AI discussions with China: notify each other when an artificial-intelligence incident could affect national security. U.S.
Micron Technology, Inc. (Micron) has filed a civil action against Yangtze Memory Technologies Co., Ltd. (YMTC) and Hongbin Zhu in the U.S. District Court for the District of Idaho. The case, filed on September 15, 2026, is listed as 1:26-cv-00604.
In a report dated October 1, Proofpoint said it observed a July 2026 credential-phishing campaign against AI-policy experts working at U.S. think tanks, universities, and legal-sector organizations. Proofpoint identified TA419, the threat actor it tracks, as China-aligned and espionage-motivated. That assessment needs careful boundaries.
