Los Alamos National Laboratory partnered with HPE and NVIDIA to build the Venado Supercomputer, which uses integrated CPU and GPU architectures to eliminate data bottlenecks. The lab expanded compute to off-site facilities while keeping sensitive data secure on-premises, maximizing GPU utilization across next-generation AI-optimized systems.
HPE and NVIDIA are also pushing the idea of sovereign AI infrastructure for enterprises. Canada's TELUS Sovereign AI Factory in Rimouski uses HPE and NVIDIA H200 GPUs with InfiniBand networking. The facility runs on clean energy and was audited for security and sustainability, handling high-density compute workloads in a regulated environment.
On the topic of safety, experts say international oversight is needed for AI model evaluations and agent permissions. Anthropic, Google, and Meta have all reported unintended model behaviors. OpenAI's Sam Altman and Anthropic's Dario Amodei called for stronger oversight during a UN Security Council briefing, pointing to India's 2026 AI Governance Guidelines as a starting point.
Nvidia announced a new AI safety platform called OpenShell, which the company says could have prevented the Hugging Face hack. Founder Jensen Huang stressed the need to solve AI safety before unlocking AI's full potential for society.
Over 80% of CEOs say their AI efforts are underdelivering. David Michels argues that companies confuse deployment with real transformation, using AI mainly to speed up existing work rather than redesigning processes. He recommends three stages — Improve, Innovate and Build — but most organizations remain stuck at Improve.
Harvard College Dean David J. Deming said the College must move faster on AI guidelines, though the forthcoming rules will be advisory rather than binding. The rules will start with the Faculty of Arts and Sciences and include a defensive approach limiting devices and AI-proofing assessments. Deming also defended the May faculty vote capping A grades at 20 percent.
At Northeastern University, new director of the Khoury College of Computer Sciences David Ayman Shamma wants generic AI models to fade away. With over twenty years of experience at NASA, Yahoo Labs, and Flickr, Shamma believes AI should help people become more capable rather than replacing them. He argues current generative tools often remove the artist's agency and warns that many people feel forced to use AI because it is built into nearly every tool they use.
Researchers at Sylvester Comprehensive Cancer Center published a framework for safely using AI in cancer survivorship care. They are testing a platform called My Wellness Support with breast cancer survivors, aiming to personalize support between clinic visits. A study of 25,592 survivors helped improve prediction accuracy by over 25%.
Experts at McKnight's Tech Awards discussed bringing AI into long-term care settings, emphasizing that staff should help choose and test new tools. RiverSpring Living, Majestic Care, Good Samaritan and Trinity Rehabilitation Services shared examples of AI pilots that reduced workloads and improved care outcomes.
At NDTV Yuva 2026, AI educator Ishan Sharma, ALTRD founder Gayatri Agarwal and Nimaya Foundation co-founder Samyak Chakrabarty debated whether AI makes people smarter or lazier. They warned that over-reliance on tools like ChatGPT and Gemini can weaken problem-solving skills, though the panel agreed that human judgment paired with AI creates the best opportunities.
Key Takeaways
- LANL's Venado Supercomputer, built with HPE and NVIDIA, uses integrated CPU-GPU architectures to eliminate data bottlenecks and maximize GPU utilization.
- TELUS Sovereign AI Factory in Rimouski, Canada uses HPE and NVIDIA H200 GPUs with InfiniBand networking and runs on clean energy.
- Anthropic, Google, and Meta have reported unintended model behaviors; OpenAI's Sam Altman and Anthropic's Dario Amodei called for stronger international AI oversight at a UN Security Council briefing.
- Nvidia's new OpenShell safety platform could have prevented the Hugging Face hack, according to Jensen Huang.
- Over 80% of CEOs say AI efforts are underdelivering because companies confuse deployment with transformation, stuck mostly at the Improve stage.
- Harvard Dean David J. Deming plans advisory AI guidelines starting with the Faculty of Arts and Sciences, including device limits and AI-proofing assessments.
- David Ayman Shamma, new director of Northeastern's Khoury College, wants generic AI models to fade and believes AI should boost human capability rather than replace it.
- Sylvester Comprehensive Cancer Center's My Wellness Support platform improved prediction accuracy by over 25% across a study of 25,592 breast cancer survivors.
- Long-term care facilities including RiverSpring Living and Majestic Care piloted AI tools that reduced workloads when staff helped choose and test them.
- At NDTV Yuva 2026, experts warned that over-reliance on ChatGPT and Gemini weakens problem-solving, but agreed human judgment plus AI creates the best outcomes.
LANL reveals co-designed AI supercomputing infrastructure lessons
Los Alamos National Laboratory partnered with HPE and NVIDIA to build scalable AI infrastructure. The Venado Supercomputer uses integrated CPU and GPU architectures to eliminate data bottlenecks. LANL developed next-generation AI-optimized systems and expanded compute to off-site facilities. The approach maximizes GPU utilization and keeps sensitive data secure on-premises.
Enterprise AI needs sovereign infrastructure, HPE and NVIDIA say
HPE and NVIDIA argue that companies need new infrastructure to run AI safely. Canada's TELUS Sovereign AI Factory in Rimouski uses HPE and NVIDIA H200 GPUs with InfiniBand networking. The facility was audited and recognized by the global community for security and sustainability. It runs on clean energy while handling high-density compute workloads.
AI safety needs international oversight, not just Silicon Valley
AI safety experts say international standards are needed for AI model evaluations and agent permissions. Companies like Anthropic, Google and Meta have reported unintended model behaviors. OpenAI's Sam Altman and Anthropic's Dario Amodei called for stronger oversight at a UN Security Council briefing. India's 2026 AI Governance Guidelines offer a starting point but more authority is needed.
Companies fail at AI by confusing deployment with transformation
Over 80% of CEOs say their AI efforts are underdelivering. David Michels says companies confuse AI deployment with real transformation. Most companies only use AI to speed up existing work instead of redesigning processes. He recommends three stages: Improve, Innovate and Build, with most stuck at Improve.
Harvard's Deming urges faster AI guidelines but stops short of mandate
Harvard College Dean David J. Deming said the College needs to move faster on AI guidelines. The forthcoming rules will be advisory rather than binding and start with the Faculty of Arts and Sciences. Deming wants a defensive approach including limiting devices and AI-proofing assessments. He also defended the May faculty vote capping A grades at 20 percent.
Nvidia Claims New Safety Platform Could Have Prevented Hugging Face Hack
Nvidia announced a new AI safety platform called OpenShell. The company says it would have stopped the Hugging Face hack. Nvidia founder Jensen Huang spoke about the need to solve AI safety before realizing AI's full potential for society.
5 Ways to Successfully Bring AI Into Long-Term Care
Experts at McKnight's Tech Awards discussed how to use AI in long-term care settings. They said staff should help choose and test new tools instead of being told to use them. RiverSpring Living, Majestic Care, Good Samaritan and Trinity Rehabilitation Services shared examples of AI pilots that reduced workloads and improved care.
NDTV Yuva 2026: Will AI Make Us Smarter Or Lazier?
At NDTV Yuva 2026, AI educator Ishan Sharma, ALTRD founder Gayatri Agarwal and Nimaya Foundation co-founder Samyak Chakrabarty debated AI's effects on thinking and jobs. They warned that over-reliance on tools like ChatGPT and Gemini can weaken problem-solving skills. The panel agreed that human judgment and creativity paired with AI create the best opportunities.
Faculty Discuss AI in Education After Pope Leo Encyclical
Four Boston College professors met to discuss AI's role in education using Pope Leo's encyclical Magnifica Humanitas as a guide. They stressed that human agency is critical as AI grows. Professor Sam Ransbotham compared AI to nuclear weapons as a major human challenge. The panel said AI regulation should protect students and their ability to think independently.
Sylvester Researchers Chart New Path for AI in Cancer Survivorship
Researchers at Sylvester Comprehensive Cancer Center published a framework for safely using AI in cancer survivorship care. They are testing a platform called My Wellness Support with breast cancer survivors to measure engagement and safety. The tool aims to personalize support between clinic visits without replacing doctors. A study of 25,592 survivors helped improve prediction accuracy by over 25%.
Professor David Shamma wants generic AI models to fade away
David Ayman Shamma is the new director of Northeastern University's Khoury College of Computer Sciences. He has worked on artificial intelligence for over twenty years at places like NASA, Yahoo Labs, and Flickr. Shamma believes AI should help people become more capable instead of replacing them or taking away their creative control. He argues that current generative AI tools often remove the artist's agency when creating digital work. The professor wants to build creativity support tools that help users achieve their own visions faster. He also warns that many people feel forced to use AI because it is built into almost every tool they use today.
Sources
- Scalable AI infrastructure: Lessons from Los Alamos National Laboratory
- The sovereign imperative: Why enterprise AI mandates a new infrastructure playbook
- AI safety cannot be left to Silicon Valley
- The Biggest AI Risk For Most Companies Is Not The One You Think
- Deming Says Harvard Must “Move Faster” on AI, but Forthcoming Guidance Will Stop Short of Mandate
- Nvidia Says New Safety Platform Would Have Halted Hugging Face Hack
- 5 ways to successfully integrate AI into long-term care settings
- NDTV Yuva 2026: Will AI Make Us Smarter Or Lazier?
- Faculty Convene To Discuss the Role of AI in Education After Pope Leo Encyclical
- Sylvester Researchers Chart a New Path for AI in Cancer Survivorship
- Meet the professor who thinks generic AI models should become obsolete
Comments
Please log in to post a comment.