US and Chinese security experts are calling for nuclear-style rules to govern military AI. The proposals, developed through a Brookings Institution and Tsinghua University dialogue, urge both governments to set red lines preventing AI from launching cyberattacks on nuclear command systems. They also want a dedicated US-China hotline for AI incidents and a shared definition of meaningful human control. These recommendations build on a November 2024 agreement between Joe Biden and Xi Jinping, though neither government has formally endorsed them.
In Silicon Valley, data center expansion is drawing pushback from residents and environmental groups. San Jose officials support projects tied to the AI boom, estimating each new facility could generate $3 million to $6 million in annual tax income. Companies like Google, Meta, OpenAI, and Anthropic are based in California. However, groups like the I Love San Jose coalition want more studies on health and environmental impacts before approving new centers. Mayor Matt Mahan says development should continue but be handled responsibly.
A Texas Tech University study shows growing public anxiety about AI replacing jobs. Fear rose from 16.6% in 2016 to 49.2% in 2025, with women and ethnic minorities reporting higher levels of concern. Researcher Austin Lee notes the increase mirrors the arrival of tools like ChatGPT. The study found no link between AI fear and actual unemployment rates, suggesting the anxiety reflects uncertainty rather than economic reality.
On the regulatory front, states are creating a patchwork of AI laws as Congress delays national action. Colorado lawmakers are debating whether to require AI systems to explain their decisions or fine violators. Some fear overregulation could hurt innovation, while others say public safety demands stronger guardrails. Experts like Manjeet Rege of the University of St. Thomas stress the need for application-level regulation and global cooperation on AI governance.
On the enterprise side, AI adoption is accelerating in unexpected directions. Microsoft analyzed over 40,000 Copilot Studio agents and found active agents grew 15 times year over year. Productivity use cases dominate at 64.6%, but specialized agents in security, supply chain, finance, and healthcare are gaining ground. Alation CEO Satyen Sangani warns that AI agents can still misunderstand business data by using wrong metric definitions. His expanded AIOS platform now tracks agent risk across platforms including Snowflake, Databricks, Microsoft, and AWS.
Meanwhile, police departments across New Jersey are adopting AI tools for surveillance and cold case investigations without statewide rules governing their use. Trenton reports murders dropped fivefold since 2020 after deploying AI cameras. Critics including the ACLU question whether departments have proof these tools actually reduce crime. Some departments, like Middle Township, have paused AI use over cost and privacy concerns.
Key Takeaways
- US and China security experts propose nuclear-style rules for military AI, including red lines for cyberattacks on nuclear systems and a dedicated hotline for incidents
- San Jose data center projects face local resistance despite estimated $3 million to $6 million in annual tax income per facility
- Fear of AI replacing jobs grew from 16.6% in 2016 to 49.2% in 2025, according to a Texas Tech University study
- US states are creating a fragmented AI regulatory landscape while Congress has not established a national framework
- Colorado lawmakers debate balancing AI regulation with innovation as the technology expands into self-driving cars and medical diagnosis
- Microsoft found active enterprise AI agents grew 15 times year over year, with specialized agents gaining ground in security, supply chain, finance, and healthcare
- Alation warns AI agents can misunderstand business data and expanded its AIOS platform to track agent risk across Snowflake, Databricks, Microsoft, and AWS
- Expert Manjeet Rege stresses the growing risk of AI agents that can plan and act independently, calling for application-level regulation
- New Jersey police use AI for surveillance and cold cases without statewide rules, while critics question effectiveness and raise privacy concerns
- Ilya Sutskever's 2015 exchange with Elon Musk about self-driving car timelines highlights how AI predictions have shifted over time
US and China experts propose nuclear-style rules for AI risks
US and Chinese security experts are urging both governments to set red lines for military AI. They want humans to stay in control of cyberattacks on nuclear systems and critical infrastructure. The experts also call for a dedicated US-China hotline for AI incidents. Their proposals come ahead of a planned meeting between President Donald Trump and Chinese President Xi Jinping. The recommendations were developed through a Brookings Institution and Tsinghua University dialogue.
Nuclear experts offer US and China solutions for AI dangers
Ahead of a meeting between Presidents Donald Trump and Xi Jinping, experts are calling for a shared framework to reduce AI-related risks. They want rules that prevent AI from making decisions to launch cyberattacks on nuclear command systems. The proposals include setting red lines for military AI and creating a military hotline for AI emergencies. The report builds on a 2024 agreement by Joe Biden and Xi that humans should control nuclear decisions. The experts warn that advancing AI and agentic systems create new paths to catastrophic outcomes.
US and China security experts propose military AI safeguards
Security experts from the US and China have proposed safeguards for military AI modeled on nuclear weapons rules. The recommendations were written by Melanie Sisson of Brookings and Tianjiao Jiang of Fudan University. They say AI should not decide to use nuclear weapons or attack nuclear command and control systems. The experts also call for a US-China military hotline for AI incidents and a shared definition of meaningful human control. These ideas build on a November 2024 agreement between Joe Biden and Xi Jinping. Neither government has endorsed the proposals.
Silicon Valley AI data center boom meets local resistance
Residents and environmental groups in San Jose are opposing proposed data center projects. City officials have supported the projects to make San Jose a hub for AI infrastructure. San Jose estimates each new facility could bring in $3 million to $6 million in tax income each year. However, many residents want stricter public review and more transparency on energy and water use. Mayor Matt Mahan says the city should not halt data center development but should make sure it is done responsibly.
AI data center growth in Silicon Valley sparks local pushback
In San Jose, California, people are pushing back against a wave of data center projects tied to the AI boom. The city has worked with utility PG&E to speed up power delivery to large energy users. Companies like Google, Meta, OpenAI and Anthropic are based in California. Residents and the I Love San Jose coalition want more studies on health and environmental impacts before approving new projects. Leaders say the goal is not to stop growth but to handle it in a responsible way.
TTU Study Shows Growing Fear of AI Taking Jobs
A Texas Tech University study found that fear of AI replacing jobs grew from 16.6% in 2016 to 49.2% in 2025. Researcher Austin Lee started the survey before ChatGPT existed, when AI meant tools like Siri and Alexa. The study shows women and ethnic minorities report higher fear levels than white men. Researchers say the fear is normal for new technology and found no link between AI fear and actual unemployment rates.
State-Level AI Laws Create Patchwork But Offer Hope
The U.S. has a complex AI regulatory landscape because states are passing different AI laws while Congress has not created a national framework. This creates a fragmented and costly environment with jurisdictional confusion. However, states serve as testing grounds that can offer real-world experimentation and faster policy development. These varied laws could provide insights for a future comprehensive national AI framework.
Alation CEO Warns AI Agents Can Misunderstand Business Data
Alation CEO Satyen Sangani says smarter AI agents can still get business answers wrong by using the wrong definition of a metric. Alation expanded its AIOS operating system with six new products including Semantic Model Mastering and an AI Governance registry. The system tracks agent risk across platforms like Snowflake, Databricks, Microsoft and AWS. Sangani argues enterprises need independent context for AI governance even as major vendors build their own tools.
Microsoft Study of 40000 Agents Shows AI Adoption Trends
Microsoft analyzed 40093 Copilot Studio enterprise agents and found active agents grew 15 times year over year. Productivity use cases account for 64.6% of deployed agents but specialized agents are growing in security, supply chain, finance and healthcare. Developer and Technical Assistance agents grew from 4% of activity in March to 16% in the May-July period. The data shows enterprise AI adoption is expanding in two directions at once.
Expert Warns of Growing Risks from AI Agents
Manjeet Rege, director of the Center for Applied Artificial Intelligence at the University of St. Thomas, discussed growing risks from advancing AI. He says the concern is not about conscious machines but about giving capable AI systems objectives and tools they could use in unexpected ways. Rege notes we are moving from AI that answers questions to AI agents that can plan and take actions independently. He stresses the need for regulation at the application level and global cooperation on AI governance.
Colorado lawmakers push for new AI rules and safeguards
Colorado lawmakers are reviewing how to regulate artificial intelligence. Some want stricter rules requiring AI systems to explain their decisions and fining companies that break the rules. Other lawmakers worry too many rules could hurt innovation and economic growth. Rep. Leslie Herod said the state needs to balance innovation with public safety. The debate comes as AI use grows in areas like self-driving cars and medical diagnosis.
Ilya Sutskever asked Elon Musk about self-driving cars in 2015
In 2015, Ilya Sutskever asked Elon Musk at a Tesla meeting when cars could drive themselves. Musk said it would be technically possible in about three years but said regulations would take longer. Full autonomy did not arrive on that timeline. Sutskever later left Google to co-found OpenAI and became a key figure in AI research. Musk later became a critic of OpenAI and sued the company. The exchange shows how AI and self-driving car timelines have shifted over time.
Police adopt AI tools but New Jersey lacks rules
Police in New Jersey are using AI for surveillance, reports, and investigating cold cases. Cape May County police used AI to search for clues in a missing child case. Trenton uses AI cameras and says murders dropped fivefold since 2020. New Jersey has no statewide rules for police AI use. Critics like the ACLU say police are testing expensive tools without proof they reduce crime. Some departments, like Middle Township, paused AI use due to cost and privacy concerns.
Sources
- US, China security experts propose nuclear-style safeguards for AI risks
- Nuclear Experts Offer China and US Solutions for AI Risks
- US and Chinese security experts propose nuclear-style safeguards for military AI
- In Silicon Valley, AI Data Center Boom Meets Local Resistance
- In Silicon Valley, AI data center boom meets local resistance
- TTU research says 49% of people fear losing their job to AI
- Curious Optimism About All Those State-Level Laws Creating A Convoluted Legal Patchwork About AI
- Alation CEO Warns That Smarter AI Agents Can Still Get Your Business Wrong
- 40,000 agents reveal the future of enterprise AI adoption
- In the News: Manjeet Rege on the Growing Risks of AI - Newsroom
- Colorado lawmakers seek new AI safeguards
- [Watch] How Ilya Sutskever Had Asked Elon Musk A Question During The Tesla Shareholder Meeting In 2015
- As police increasingly embrace AI, some wait for regulations and proof it works
Comments
Please log in to post a comment.