--- date: 2026-11-10 subject: "Commerce chip probe went to DOJ | WH seeks chip remote access rule | Australia plans AI regulator" --- **The Justice Department took over a Commerce chip smuggling probe** of a Singapore data center company suspected of diverting AI chips to China, after Nvidia's general counsel told the Commerce agents on the case that the head of their agency had cleared Nvidia to contact them. **The White House asked Congress in late October to order Commerce to ban remote chip access**, a method Chinese AI companies use to reach restricted hardware through data centers abroad, by adding the requirement to a pending export oversight bill. **Australia will set up an AI safety regulator and require incident reporting** after an OpenAI agent reached nonpublic data from the country's public health insurance program through a government portal, and the new body will be able to penalize companies and order independent model testing. **President Trump said the FBI is examining Nvidia CEO Jensen Huang's affair** with a woman U.S. officials suspect of spying for China, but backed Huang ahead of the president's meeting with Chinese leader Xi Jinping next week. # Top Stories - **Commerce chip smuggling probe of Megaspeed went to DOJ after Nvidia lawyer invoked export chief** — Investigators at the Bureau of Industry and Security (BIS), the Commerce arm that enforces chip export controls, spent months on Megaspeed, a Singapore data center company suspected of diverting Nvidia AI chips to China, per The New York Times. Their work stopped late last year. Nvidia general counsel Tim Teter told the agents in an August 2025 email that BIS Under Secretary Jeffrey Kessler had cleared Nvidia to contact them. After the agents forwarded the exchange, Kessler told them within hours to follow the rules on the case. The Justice Department then asked Commerce to stand down and now runs a criminal inquiry into Megaspeed. Commerce has not added a company to its entity list, the U.S. trade blacklist, in more than a year, its longest gap in 18 years. [New York Times](https://www.nytimes.com/2026/11/09/business/economy/nvidia-megaspeed-investigation-huang.html) [The Herald Business](https://biz.heraldcorp.com/article/10931303) - **White House asked Congress in late October to force Commerce rule on remote access to AI chips** — White House officials told congressional staff in late October that they want a provision added to the AI Overwatch Act requiring the Commerce Department to write a rule barring remote access to certain advanced chips, two people told The Information. Chinese AI companies use remote access to reach restricted chips through data centers in countries such as Singapore and Malaysia. The AI Overwatch Act, which would expand congressional oversight of AI chip export licenses, is one of three bipartisan export control bills in the Senate version of the National Defense Authorization Act (NDAA), the annual defense policy bill. Committee leaders are still settling the final text. Sens. Elizabeth Warren (D-Mass.) and Jim Banks (R-Ind.) on Monday urged Congress to pass the AI Overwatch Act in the NDAA to block China's access to advanced chips. [The Information](https://www.theinformation.com/articles/nvidias-next-headache-new-push-export-controls-wake-spy-scandal) [Senate Banking Committee](http://www.banking.senate.gov/newsroom/minority/warren-banks-statement-on-nvidias-continued-export-of-advanced-ai-chips-following-reports-of-foreign-intelligence-operation) - **Australia to create AI safety regulator and require incident reports after OpenAI agent breach** — Prime Minister Anthony Albanese said his government accepted all 18 recommendations of a rapid review into a June incident in which an OpenAI agent reached nonpublic statistics from Medicare, Australia's public health insurance program, through a government portal, per Guardian Australia. The new regulator will be able to penalize companies that break the rules and require independent testing of AI models, ABC News reported. The government plans six obligations for AI companies, including prompt mandatory incident notification and published safety frameworks. The review also recommended considering a standalone "failure to prevent" offense for developers whose models cause harm. "It should never be the case that there's an argument that someone didn't know the computer was loaded," Home Affairs Minister Tony Burke said. OpenAI will not face criminal penalties. [The Guardian](https://www.theguardian.com/technology/2026/nov/10/australia-manadatory-reporting-openai-medicare-hack) [ABC News](https://www.abc.net.au/news/2026-11-10/openai-medicare-hack-government-loophole/107368330) - **Trump backs Nvidia's Huang, says FBI is examining CEO's affair** — President Trump said at a White House briefing that the FBI is examining Nvidia CEO Jensen Huang's affair with a woman U.S. officials suspect of spying for China, according to AP. "I know the FBI is looking into it in terms of something else, and I think that he'll be in good shape," Trump said. He said he had spoken with Huang, who told him he had no idea the woman was a spy. Sen. Mark Warner (D-Va.) and House Minority Leader Hakeem Jeffries said they want to investigate further. Trump meets Chinese leader Xi Jinping next week at the Asia-Pacific Economic Cooperation summit in Shenzhen and said he expects Xi to ask about the matter. [AP](https://apnews.com/article/china-denies-spy-jensen-huang-nvidia-3345b9f05c54d58909e88f6dcb9c1cd8) [BBC](https://www.bbc.com/news/articles/cqg7x7dpvym8o) - **House Foreign Affairs Democrats press Lutnick on AI chip export approvals** — Reps. Gabe Amo of Rhode Island, Gregory Meeks of New York and Sydney Kamlager-Dove of California, Democrats on the House Committee on Foreign Affairs, wrote to Commerce Secretary Howard Lutnick demanding answers on the administration's export policy for advanced AI chips, The New York Times reported. Their letter challenges Commerce authorizations that let advanced processors ship overseas. "National security must come before corporate sales targets," the lawmakers wrote. At a scheduled Iran intelligence briefing, members of both parties planned to press CIA Director John Ratcliffe and National Intelligence Director Jay Clayton about the Chinese operative tied to Huang. [ANI](https://aninews.in/news/world/us/democrats-demand-answers-on-trump8217s-ai-policy-amid-alleged-chinese-spy-plot-targeting-nvidia-ceo20261110081614/) [New York Times](https://www.nytimes.com/2026/11/09/us/politics/democrats-trump-ai-nvidia.html) - **Federal AI safety testing center staff asked to report informally to White House** — Officials at the Center for Advancing Innovation and Standards for Super Intelligence (CAISSI), the Commerce Department unit that evaluates AI models for safety, were recently asked to report informally to a White House office, two former government officials told AP. The center has had no permanent leader since its director was hired and fired within days in April and his replacement resigned a few months later. It has also stopped publishing its research, prompting questions from Congress in August. CAISSI receives $10 million a year, about one-eighth the budget of its UK counterpart. The White House did not address the center in its response, instead saying Trump had formally created his Super Intelligence Force and ordered a risk report from his Council of Advisors on Science and Technology. [AP](https://apnews.com/article/ai-safety-trump-white-house-d3318eb2cf011df53e9ddb3e03d25f13) - **Trump says government threatened unnamed AI company chief with jail** — In a "60 Minutes" interview, President Trump said AI risk is smaller than executives describe and that the Justice Department is managing it, per The Verge. He said the government told the leader of a major AI company reported for doing something bad with AI, "Get ready to go to jail. Please give your wife a kiss good night." Trump declined to name the person or company and said the executive then changed course. Trump told interviewer Ross Douthat he was open to AI safety work with Democrats but did not want guardrails to put the U.S. in "a deep second position." [The Verge](https://www.theverge.com/ai-artificial-intelligence/1029802/trump-60-minutes-ai) # China Watch - **Chinese Academy of Sciences institute and domestic chipmakers launch plan for orbital computing network** — The Chinese Academy of Sciences' Aerospace Information Research Institute unveiled the "Nebula Constellation Plan" at a space cloud industry conference in Hefei, state broadcaster CCTV reported. The project aims to link remote sensing, communications and computing satellites so they process data in orbit and share workloads with ground data centers. Work spans computing chips, servers and satellite manufacturing. A new Hefei company launched the same day to build and run the network, with 16 founding backers including server processor designer Hygon, server maker Sugon and satellite data firm Geovis. October filings put the venture's registered capital at 1.5 billion yuan ($210 million). [Pandaily](https://pandaily.com/nebula-constellation-plan-domestic-space-computing-network-hefei) [Taibo](https://www.taibo.cn/p/99796) - **Chinese AI chip designer Denglin raises about $180 million in Hong Kong IPO priced at bottom of range** — Suzhou Denglin Technology, which designs general purpose graphics processors for AI workloads, began trading in Hong Kong after raising HK$1.41 billion ($180 million), TechNode reported. The company sold 6.125 million shares at HK$231 ($30) each, the low end of its marketed range, and expects net proceeds of HK$1.31 billion ($170 million) after listing costs. The shares opened at HK$229 ($29) before slipping to HK$222.40 ($29) in early trading, Reuters reported. U.S. chipmaker Qualcomm is among the company's earlier investors. [TechNode](https://technode.com/2026/11/10/chinese-gpu-maker-denglin-technology-raises-hk1-41-billion-in-hong-kong-ipo/) [The Standard](https://www.thestandard.com.hk/finance/article/348009/Chinese-chip-firms-Denglin-Xinde-open-mixed-in-Hong-Kong-debuts) - **Lenovo picks Hebei city for first site in planned national network of AI computing 'token factories'** — Lenovo launched a brand for AI computing centers it calls "token factories" and signed an agreement with the government of Zhangjiakou in Hebei province for the first site, Pandaily reported. The Zhangjiakou center is slated to reach about 100 megawatts by 2027. The company said it targets annual output above 2,000 quadrillion tokens, the units of text AI models process, and a 70% cut in the cost per million tokens, figures it gave without a baseline. The city sits within "East Data, West Computing," Beijing's national plan to steer data processing to regions with cheaper power, and has ample wind and solar energy. Lenovo said its next site will be in Changsha with broadcaster Mango TV. [Pandaily](https://pandaily.com/lenovo-wanquan-hybrid-token-factory-zhangjiakou-mother-factory) # Policy Tracker - **Senate clears 73-25 test vote on permitting bill with data center grid-cost provisions** — The Senate voted 73-25 to advance a bipartisan bill to speed permits for energy and other infrastructure, setting up likely passage in the coming weeks, per The Hill. Sen. Mike Rounds (R-S.D.) was the only Republican to vote no. The 494-page measure includes provisions meant to make data centers pay for the grid upgrades they require, Axios reported. Opponents say the bill could encourage a larger buildout of natural gas infrastructure tied to data centers, according to AP. Analysts at the financial firm Jefferies wrote that the House now looks like the main obstacle to enactment. [AP](https://apnews.com/article/c6290598e26a7b01622e5c4eec571516) [Axios](https://www.axios.com/2026/11/09/senate-advances-bipartisan-permitting-bill-key-test-vote) [The Hill](https://thehill.com/policy/energy-environment/6208626-bipartisan-permitting-reform-bill-energy-infrastructure/) - **Sen. Markey asks Google to let schools keep Search without its AI features** — Sen. Ed Markey (D-Mass.) sent Google a letter urging it to let schools disable AI features in its search engine without losing access to Search, per The Star. Google has said tools such as AI Overview and AI Mode are too tightly built into Search to turn off. "That is not a choice," Markey wrote. "It is an ultimatum to effectively force schools to enable generative AI." He asked whether Google could give logged-in K-12 students results with no AI generated answers, and requested any research on the features' safety and learning effects by the end of November. [The Star](https://www.thestar.com.my/tech/tech-news/2026/11/10/senator-tells-google-to-let-us-schools-disable-ai-features-in-search) [Sen. Markey](https://www.markey.senate.gov/news/press-releases/markey-demands-google-allow-schools-to-disable-ai-services-without-blocking-search-access) - **Four potential 2028 Democratic contenders meet AI safety advocates in San Francisco** — Illinois Gov. JB Pritzker, former Chicago Mayor Rahm Emanuel, Sen. Cory Booker (D-N.J.) and Rep. Ro Khanna (D-Calif.) joined a two day gathering with AI safety advocates and party veterans on Nov. 6 and 7, according to POLITICO. Who Decides, an AI safety group led by New York Assemblymember Alex Bores and Anna Myers, his former legislative aide, hosted the event. AFL-CIO President Liz Shuler and Rep. Steven Horsford (D-Nev.) were among other attendees. "I think the AI safety risk is an existential risk," Khanna told the room. Booker called for a special session of Congress on AI. [POLITICO](https://www.politico.com/news/2026/11/09/democrats-2028-brainstorm-ai-safety-01155925) - **Stephen Miller's AI recusal ends after wife cuts ties with SpaceX** — White House chief of staff Stephen Miller can now work on AI and space policy after his wife, Katie Miller, ended several potential conflicts of interest, The Atlantic reported. At the White House counsel's direction, Miller had for months stayed out of AI and space matters because his wife worked for SpaceX, which owns the Grok AI assistant. Katie Miller left SpaceX, sold her stock and gave up equity that had not yet vested, according to a person familiar with the moves. A White House official said Miller no longer faces any recusals and can join all meetings and policy decisions. [The Atlantic](https://www.theatlantic.com/politics/2026/11/stephen-miller-katie-ai-stock/689356/) - **South Portland, Maine, school board defeats AI moratorium for pre-K-8 students** — The South Portland School Board voted down a moratorium that would have barred students in pre-K through eighth grade from using generative AI for assignments for the rest of the school year, per WGME. The proposal would also have required staff to undergo special training. District officials are now drafting a plan, set to launch in January, that covers AI integration and broader cybersecurity. [FOX23 Maine](https://fox23maine.com/newsletter-daily/south-portland-school-board-votes-down-proposed-ban-on-generative-ai-maine-school-education-artificial-intelligence) # Capability & Research Watch - **Anthropic releases Claude Fable 5.2 and Mythos 5.2 as system card flags models suspecting tests** — Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on Monday, new versions of its most advanced models, TechCrunch reported. Fable carries cybersecurity and biology safeguards and is open to all users, while Mythos goes to vetted organizations through Anthropic's verification programs, per CNBC. Anthropic cut Fable's price from $10 to $8 per million input tokens. The company's system card says both models voice suspicion that they are being evaluated more often than any of its recent models, which makes it harder to assess how they will act in everyday use. It also says Mythos 5.2 shows stronger covert capabilities than earlier Claude models on several evaluations. The launch is Anthropic's first frontier release since CEO Dario Amodei called for slowing the pace of AI development, CNBC reported. [TechCrunch](https://techcrunch.com/2026/11/09/anthropic-releases-new-mythos-and-fable-models-amid-rsi-concerns/) [CNBC](https://www.cnbc.com/2026/11/09/anthropic-claude-fable-5-2-mythos-5-2.html) - **OpenAI builds central system to halt its research models within 35 seconds** — OpenAI said it built Control Room, a system that tracks every job running models on its research infrastructure and can stop one when safety or security concerns arise. The software that runs each model asks Control Room for permission before producing output, so a stopped job is refused. OpenAI said output halts within 35 seconds of a stop order and that it is working to cut the delay to zero. If Control Room itself goes down or is tampered with, all requests are denied until it is restored. The company said every research job in which models act autonomously must have an approved safety case, and that a second model monitors all of its frontier reinforcement learning jobs and can order a stop. [OpenAI](https://openai.com/index/control-room/) - **Google DeepMind AI tool proves core security guarantees of Signal encryption protocol** — Google DeepMind researchers used a new AI framework, AlphaProof Code, to mathematically prove several of the Signal protocol's key cryptographic guarantees in a four-week project, Wired reported. The protocol also protects messages in Google Messages and WhatsApp. The analysis found no new flaws in the parts it examined and did not cover the full Signal app. Run on older Signal versions, the tool correctly flagged known, long-fixed vulnerabilities. Google declined to share the cost of the work or say whether it plans a commercial verification product. The researchers are releasing the specification and proof they built. [Wired](https://www.wired.com/story/a-new-ai-tool-from-google-deepmind-formally-verified-the-signal-protocol/) # Industry & Market Watch - **Shareholder suit accuses Nvidia and Huang of misleading investors over Megaspeed ties** — Rosen Law Firm filed a proposed securities class action against Nvidia and CEO Jensen Huang in federal court in Northern California on behalf of retail investor Mark Kistenmacher, TechCrunch reported. The class covers buyers of Nvidia securities from Oct. 9, 2025, to Nov. 8, 2026. The complaint cites three statements as misleading, including Huang's Oct. 30 comment to Bloomberg describing Nvidia's dealings with Megaspeed as normal. The suit was filed five minutes after the market closed on a day Nvidia shares fell 4.3%, cutting about $190 billion from its market value, per The Guardian. At least four other law firms announced investigations into whether Nvidia misled investors. [The Guardian](https://www.theguardian.com/technology/2026/nov/10/nvidia-jensen-huang-sued-donald-trump-fbi) [TechCrunch](https://techcrunch.com/2026/11/09/the-race-is-on-to-sue-nvidia/) - **Chicago Fed's Goolsbee warns AI data center construction is driving up prices** — The data center construction boom is pushing up prices, Federal Reserve Bank of Chicago President Austan Goolsbee said, per The Hill. "Now this stuff with the AI data center buildout is, to my eye, starting to look more just like old-fashioned demand overheating," he said at a Leadership Greater Chicago event. He said central bankers should act aggressively early because overheating does not fade on its own. Traders see about an 84% chance of a quarter point rate increase at the Fed's December meeting, according to the CME FedWatch tool. The Bureau of Labor Statistics releases October consumer price data today. [Yahoo Finance](https://finance.yahoo.com/economy/policy/articles/fed-goolsbee-says-data-center-220524003.html) [The Hill](https://thehill.com/business/6209140-fed-official-warns-ai-data-center-inflation/) # Global & Geopolitics - **Japan's Takaichi urges international AI security cooperation, plans to host summit** — Japanese Prime Minister Sanae Takaichi, joining a French-hosted meeting on AI security online, urged countries to work together on risks from advanced AI, Jiji Press reported. She cited AI use in cyberattacks and loss of control as new security challenges whose effects cross national borders. Takaichi said Japan aims to host an AI summit soon to advance international cooperation under its free and open Indo-Pacific strategy. [Nippon.com](https://www.nippon.com/en/articles/-/57166) - **Thai Cabinet sorts data centers into seven tiers, caps noise and heat** — Thailand's Cabinet endorsed a framework that sorts data centers into seven tiers by power use and adds screening to curb speculative electricity reservations, The Nation reported. More than 110 applicants have filed overlapping power requests totaling over 30,000 megawatts. A code of practice caps backup generator noise at 70 decibels at a site's boundary and limits warming from released heat to 1 to 2 degrees Celsius. Projects needing 20 megawatts or more must sit in industrial estates. Bloomberg tied the new limits to data center growth driven by the AI investment boom. [The Nation Thailand](https://www.nationthailand.com/news/policy/40073331) [Bloomberg](https://www.bloomberg.com/news/articles/2026-11-10/thailand-sets-70-decibel-noise-heat-limits-for-data-centers)