Nvidia CEO Jensen Huang has sparked debate by saying AI does not need new safety laws. He compared AI to the internet, arguing it has thrived for decades without fresh regulations. Instead, Huang favors education and awareness over government rules. Some current and former Nvidia employees have criticized Huang and expressed concern about AI risks. Huang also says AI firm leaders themselves should decide whether to release new versions of their systems.
Meanwhile, political voices are weighing in on AI oversight. Senator Bernie Sanders criticized former President Donald Trump for what he called a lack of AI understanding during a Senate hearing. Sanders announced legislation to ban superintelligent AI that could surpass human intelligence, saying Congress has been asleep at the wheel on oversight. A former U.S. lawmaker also warned in Duncan that AI will never be fully safe, though developers and users can work to reduce risks.
On the research front, a study found safety signals that could help test AI agents in datacenter networks. Researchers built a ground truth to evaluate Agentic Network Operations agents during network repair tasks. Tests across 10 agent models showed that internal signals predict harm and progress better than observable ones. The team hopes these signals will help agents avoid risky actions.
In healthcare, AI is showing both promise and pitfalls. An AI tool from Aidoc found 39% more brain aneurysms that radiologists missed in a study of 3,856 CT scans. It correctly identified 84.6% of true aneurysms versus 71.8% for radiologists, but flagged more than one in five positives as false. Knowtex also launched a frontier healthcare AI lab to test clinical AI tools for accuracy, auditability and clinician trust, aiming to improve patient outcomes and cut costs.
Elsewhere, Ark senior executive Wood praised a global AI leaders' pledge, calling it a positive step toward responsible and transparent AI development. Iowa is grappling with surging electricity demand from AI data centers, with researchers from the University of Iowa, Iowa State University and Drake University urging the state to expand wind and solar power rather than build new fossil fuel plants. IIT Kanpur also announced a five-week online Python AI and ML course starting December 14, 2026, running through January 15, 2027.
Key Takeaways
- <ul><li>Jensen Huang says AI does not need new safety laws and favors education over regulation</li><li>Some current and former Nvidia employees have criticized Huang and raised AI risk concerns</li><li>Huang says AI firm leaders should decide whether to release new AI versions</li><li>Bernie Sanders criticized former President Trump on AI understanding and announced legislation to ban superintelligent AI</li><li>A former U.S. lawmaker warned AI will never be fully safe but can be made safer</li><li>Study of 10 agent models found internal signals predict harm better than observable ones in datacenter networks</li><li>Aidoc AI found 39% more brain aneurysms in 3,856 CT scans but flagged over 21% as false positives</li><li>Knowtex launched a frontier healthcare AI lab focused on accuracy, auditability and clinician trust</li><li>Ark's Wood praised a global AI leaders' pledge for responsible and transparent development</li><li>Iowa researchers urge wind and solar expansion as AI data centers strain the energy grid</li></ul>
Nvidia CEO says AI does not need new safety laws
Jensen Huang, the CEO of Nvidia, says AI does not need new laws to be safe. He compared AI to the internet, which has worked for decades without new rules. Some current and former Nvidia employees worry about AI risks. Huang wants more education and awareness about AI instead of new regulations.
Nvidia boss says AI leaders should decide on new versions
Nvidia boss Jensen Huang says the heads of AI firms should decide if new versions of AI should be released. His comments come as concerns grow about AI's threat to humans. Huang is a leader of the recent AI boom. The article does not give more details about his specific suggestions.
Study finds safety signals to test network AI agents
Researchers built a ground truth to test Agentic Network Operations agents in datacenter networks. They measured whether each action makes progress or causes harm during network repair tasks. Tests across 10 agent models showed that internal signals predict harm and progress better than observable signals. The team aims to use these signals to help agents avoid risky actions.
Ark's Wood praises global AI leaders' pledge
Ark's Wood, a senior executive at Ark, said a global AI leaders' pledge is positive for technology development. The pledge is a commitment by global AI leaders to develop and implement AI responsibly and transparently. Wood said the pledge will help ensure AI benefits society. He called it a step in the right direction.
AI finds more brain aneurysms but also more false alarms
An AI tool from Aidoc found 39% more brain aneurysms that radiologists missed in a study of 3,856 CT scans. The AI correctly found 84.6% of true aneurysms compared with 71.8% for radiologists. However, the AI had a lower accuracy rate, flagging more than one in five positives as false. The study was published in the Journal of the American College of Radiology.
Former lawmaker warns AI will never be fully safe
A former U.S. lawmaker spoke in Duncan about artificial intelligence and ethical concerns. He said AI will never be completely safe but developers and users can work to make it safer. The expert focused on national security, energy policy and American politics.
IIT Kanpur launches 5-week Python AI and ML course
IIT Kanpur announced a five-week online certification program on Python for AI, ML and deep learning starting December 14, 2026. The course runs through January 15, 2027 on Zoom and is open to students, faculty and professionals. Participants will use tools like TensorFlow, Keras and Pandas on real-world datasets. Registration windows include a golden deadline of October 4, 2026 and spot registration on December 15, 2026.
AI data centers push Iowa to a energy crossroads
Surging electricity demand from AI data centers is forcing Iowa to choose its energy future. The Iowa Climate Statement 2026 was signed by researchers from the University of Iowa, Iowa State University and Drake University. Iowa ranks among the top five states for data centers per capita. Researchers say Iowa should expand wind and solar power instead of building new fossil fuel plants.
Bernie Sanders criticizes Trump on AI understanding
Senator Bernie Sanders criticized former President Donald Trump for lacking understanding of artificial intelligence. Sanders called Trump's AI ignorance truly embarrassing during a Senate hearing. He also announced legislation to ban superintelligent AI that could surpass human intelligence. Sanders said Congress has been asleep at the wheel on AI oversight.
Knowtex opens frontier healthcare AI lab
Knowtex launched a frontier healthcare AI lab focused on developing clinical AI tools. The lab will test tools for accuracy, auditability and clinician trust. It aims to improve patient outcomes and reduce healthcare costs. Knowtex will work with healthcare providers and researchers to build reliable AI tools.
Sources
- Nvidia boss says AI 'doesn't need new laws' as safety concerns grow
- Nvidia boss says AI 'doesn't need new laws' as safety concerns grow
- Safety Signals to Verify NetOps Agents with Action-Level Granularity
- Ark’s Wood says global AI leaders’ pledge is positive for technology development
- AI Boosts Brain Aneurysm Detection but Produces More False Positives
- Former lawmaker discusses AI, ethical issues
- IIT Kanpur announces 5-week online certificate course on Python, AI, ML
- Surging electricity demand from AI data centers puts Iowa at an energy crossroads
- Trump's AI ignorance is 'truly embarrassing,' Bernie Sanders says - Live Updates
- Knowtex launches frontier healthcare AI lab
Comments
Please log in to post a comment.