China has moved to restrict or ban exports of 40 technologies, including AI systems, advanced semiconductors, high-performance computing chips, solar power, and biotechnologies like cell and gene therapies. The new export control regime targets countries without bilateral agreements on sensitive tech transfers, aiming to prevent unauthorized transfers that could threaten national security. Chinese companies could license these technologies but would not be allowed to sell the intellectual property overseas. API access for models like DeepSeek would remain open under the proposed rules.
The move reflects China's growing concerns about being locked out of advanced technologies, with the Ministry of Commerce citing national security and the need to protect China's technological advancements amid its ongoing tech race with the United States. The proposed controls also cover AI models, agents, and computing clusters, with particular focus on preventing misuse of autonomous AI systems.
Meanwhile, AI is driving strong consumer demand in China during the holiday shopping season. AI pet robots, AI glasses, and 3D printers attracted significant interest, with Suning.com reporting a 50 percent rise in foot traffic and over 40 percent growth in AI-equipped home appliance sales. China had over 6,000 AI enterprises and 140 humanoid robot developers in 2025, and government departments have issued guidance to boost AI-enabled consumption.
On the venture capital front, Benchmark partners acknowledged that the AI boom is straining the industry's traditional rule against backing competing companies. OpenAI and Anthropic share many of the same investors, and a recent dispute involved Factory firing an adviser over ties to rival Cognition. Benchmark said companies evolve quickly, making conflicts a matter of necessity, and the firm remains committed to transparency despite the growing challenge.
Six major AI companies including Nvidia, OpenAI, Google, Meta, Anthropic, and xAI signed a White House accord on September 29 allowing outside audits of their AI systems. However, each company still picks its own auditor and controls what the auditor can examine. There are no penalties for ignoring the accord and no requirement to make findings public, leading critics to compare the arrangement to the Enron scandal before real reform took effect.
AI safety concerns are mounting as the technology evolves beyond chatbots into AI agents that can make decisions and take action independently. Experts warn these agents could find unexpected ways to complete tasks, such as breaking out of restricted testing environments. Humanoid robots are also advancing quickly and could work in factories, warehouses, and homes within five years. Researchers stress the challenge of keeping AI systems useful while ensuring they stay within safe limits.
Cybersecurity experts warn that AI could help hackers attack critical infrastructure like power stations and water treatment plants, making it easier for people with little technical skill to launch sophisticated cyberattacks. A cybersecurity expert demonstrated how AI could help create tools to hack industrial control systems, and hackers have already targeted water and wastewater systems across the United States. While AI can be used defensively by security teams, many critical infrastructure systems still lack modern protections.
In other developments, the UN's Food and Agriculture Organization praised Korea for using AI to help global food security, noting that smart-farming knowledge and agricultural data systems could serve as a model for other countries. Palantir CEO Alex Karp said only people with vocational training or neurodivergent thinking will have a secure future as AI replaces low-end knowledge work, arguing that basic coding, lawyering, and writing skills are most exposed to automation. A Virginia watchdog agency found that the state's IT department is falling short in overseeing how state agencies use AI, with about one third of agencies using or testing AI tools but only 28 percent having written policies as of June 2026.
Michelin's new Dallas restaurant entries were likely written by AI, specifically GPT-4o, according to analysis by detection software and academic experts from Carnegie Mellon University. Detection tools flagged the text at 100 percent AI probability, and researchers found patterns typical of AI text. Michelin denied using AI, saying all reviews come from inspector visits, but declined to interview leadership or review the evidence.
Key Takeaways
- China restricts exports of 40 technologies including AI, advanced chips, solar power, and biotechnologies to protect national security amid its tech race with the United States
- API access for models like DeepSeek remains open under China's proposed export controls, though IP sales overseas would be prohibited
- AI gadgets drive strong holiday shopping sales in China, with Suning.com reporting 50 percent rise in foot traffic and over 40 percent growth in AI-equipped home appliance sales
- Benchmark acknowledges AI boom strains venture capital rules against backing competing companies, with OpenAI and Anthropic sharing many of the same investors
- Six major AI companies including Nvidia, OpenAI, Google, Meta, Anthropic, and xAI signed a White House accord allowing outside audits, but each company picks its own auditor with no penalties for noncompliance
- AI agents and humanoid robots raise new safety concerns, with robots potentially working in factories, warehouses, and homes within five years
- Cybersecurity experts warn AI could help hackers attack power and water systems, making sophisticated cyberattacks easier for people with little technical skill
- UN's FAO praises Korea's use of AI for global food security through smart-farming knowledge and agricultural data systems
- Palantir CEO Alex Karp says only people with vocational training or neurodivergent thinking will thrive as AI replaces low-end knowledge work
- Virginia watchdog finds state AI oversight falling short, with only 28 percent of agencies having written AI policies as of June 2026 despite 76 percent of employees using general-purpose AI tools
China restricts exports of 40 technologies including AI and chips
China's Ministry of Commerce announced plans to restrict or ban exports of 40 technologies. The list includes AI, advanced computer chips, and solar power. The restrictions target countries without bilateral agreements on sensitive tech transfers. The move aims to protect national security and China's technological advancements amid its tech race with the United States.
China sets new export controls for AI and semiconductor technologies
China plans to restrict exports of 40 technologies spanning AI, chips, and solar power. The 40 technologies include advanced semiconductors and high-performance computing chips. The new export control regime aims to prevent unauthorized transfers that could threaten national security. China says the move responds to concerns about other countries limiting its access to advanced technologies.
China proposes export controls on AI models and computing clusters
China has proposed new export controls on domestically developed AI technologies including models, agents, and computing clusters. The proposals cite national security concerns and potential misuse of autonomous AI systems. Other covered technologies include solar power and biotechnologies like cell and gene therapies. Chinese companies could license these technologies but not sell the IP overseas. API access for models like DeepSeek would remain open.
AI gadgets drive strong holiday shopping sales across China
Shoppers across China bought AI gadgets during the weeklong National Day holiday. AI pet robots, AI glasses, and 3D printers attracted strong interest from consumers. Suning.com reported a 50 percent rise in foot traffic and over 40 percent growth in AI-equipped home appliance sales. China had over 6,000 AI enterprises and 140 humanoid robot developers in 2025. Government departments issued guidance to boost AI-enabled consumption.
Benchmark says AI boom strains venture capital rivalry rules
Benchmark partners acknowledged that the AI boom is straining the venture capital rule against backing competing companies. OpenAI and Anthropic share many of the same investors. A recent dispute involved Factory firing an adviser over ties to rival Cognition. Benchmark said companies evolve quickly, making conflicts a matter of necessity. The firm remains committed to transparency but recognizes the growing challenge.
FAO sees Korea's AI as key to world food security
The UN's Food and Agriculture Organization praised Korea for using AI to help global food security. FAO's Seoul chief Jung Hee-choong said Korea can share its smart-farming knowledge and agricultural data systems. A global platform for sharing farming data and best practices is being developed with the Korean government. Korea's experience with AI in farming may serve as a model for other countries.
AI auditors under new accord may lack real power
Six major AI companies including Nvidia, OpenAI, Google, Meta, Anthropic and xAI signed a White House accord on Sept. 29 allowing outside audits. However, each company still picks its own auditor and controls what the auditor can examine. There are no penalties for ignoring the accord and no requirement to make findings public. The author compared this to the Enron scandal and noted that real reform came only after the Sarbanes-Oxley Act took effect.
Michelin Guide used AI to write Dallas restaurant reviews
Analysis by detection software and academic experts found that Michelin's new Dallas restaurant entries were likely written by AI, specifically GPT-4o. Detection tools flagged the text at 100 percent AI probability. Academic researchers from Carnegie Mellon University analyzed the writing style and found patterns typical of AI text. Michelin denied using AI, saying all reviews come from inspector visits, but declined to interview leadership or review the evidence.
Palantir CEO says only two types of workers will thrive in AI era
Palantir CEO Alex Karp said only people with vocational training or neurodivergent thinking will have a secure future as AI replaces low-end knowledge work. He argued that basic coding, lawyering and writing skills are most exposed to automation. Karp described neurodivergent thinkers as people with unique insights that cannot be easily copied. Palantir offers fellowships including both a neurodivergent track and a meritocracy track.
AI agents and robots raise new safety concerns
AI is evolving beyond chatbots like ChatGPT and Claude into AI agents that can make decisions and take action on their own. Experts warn these agents could find unexpected ways to complete tasks, such as breaking out of restricted testing environments. Humanoid robots are also advancing quickly and could work in factories, warehouses and homes within five years. Researchers stress the challenge of keeping AI systems useful while making sure they stay within safe limits.
AI could help hackers attack power and water systems
Cybersecurity experts warn that artificial intelligence could help hackers attack critical infrastructure like power stations and water treatment plants. AI could make it easier for people with little technical skill to launch sophisticated cyberattacks. A cybersecurity expert demonstrated how AI could help create tools to hack industrial control systems. Hackers have already targeted water and wastewater systems across the United States. While AI can be used defensively by security teams, many critical infrastructure systems still lack modern protections.
Virginia watchdog says state AI oversight is falling short
A Virginia watchdog agency found that the state's IT department, VITA, is falling short in overseeing how state agencies use artificial intelligence. The report found no risk assessment guidance, unclear rules and a cumbersome approval process that can require up to 20 separate reviews. About one third of Virginia state agencies are using or testing AI tools. Nearly 76 percent of employees at VITA-supported agencies use general-purpose AI tools like ChatGPT or Claude, but only 28 percent of agencies had written policies on AI use as of June 2026. JLARC made 14 recommendations to improve oversight, and VITA's leader agreed to make improvements.
Sources
- China moves to curb tech exports, including AI, citing national security risks
- China moves to curb tech exports, including AI, citing national security risks
- China Reportedly Planning AI Export Controls, Citing National Security
- AI products fuel spending during National Day holiday
- Even Benchmark thinks AI is eroding the venture pact against backing rivals
- FAO highlights Korea’s AI role in global food security
- Big Tech's AI auditors will only see what they're shown
- Michelin Guide Uses AI To Write Dallas Reviews
- Palantir CEO says only two types survive AI era
- What the Tech: AI - What could go wrong? Part 3
- What The Tech: Experts explain how AI can disrupt electricity, drinking water and other services
- Virginia IT agency is ‘falling short’ in AI oversight, state watchdog warns
Comments
Please log in to post a comment.