NVIDIA is pushing for AI security to be treated as a proper engineering discipline, not just a research concern. The company released a blueprint for securing AI agent stacks across the full technology stack, calling for defined security requirements, named owners, and measurable evidence that protections actually work. NVIDIA argues that established security practices from the internet and cloud era can be adapted for AI, and that smaller organizations need broader access to defensive tools.
Meanwhile, UK security leaders are sounding the alarm. A survey of 100 UK CISOs found that 59% believe attackers currently have the advantage because of AI. Many organizations still rely on manual security processes that take over a week to resolve critical issues, and only 33% of leaders feel very prepared for AI-driven attacks. Lack of trust and governance concerns remain the biggest barriers to faster automation.
On the policy front, former President Barack Obama called for stronger federal regulation of AI during a speech at Colgate University on September 18. He warned that companies' profit goals may not align with public safety and said Congress needs to catch up with the pace of AI development. Days later, President Donald Trump announced plans for an AI Force modeled after the Space Force and asked the public to help rename artificial intelligence, with options like Superior Intelligence, Extreme Intelligence, and Supreme Intelligence.
Education and youth access are also in focus. California's new law limiting AI use in education is being praised as a positive step, with experts urging schools to set their own limits and teach children about responsible AI use. A recent study found that 40% of children under 12 have used voice assistants like Alexa or Google Home, and 60% have tried virtual reality headsets. Parents are encouraged to set clear rules and model responsible technology behavior.
OpenAI is taking a different approach by forming an independent Advisory Group on Mathematics and Artificial Intelligence. The group will guide how new AI research results are reviewed and shared, connecting mathematicians with OpenAI on academic standards. Members are not paid by OpenAI and the group can make its views public. In parallel, Latvia launched a free AI skills training program through Riga TechGirls, offering courses on Coursera to over 1,000 participants using tools like Gemini, NotebookLM, and Google AI Studio, supported by the Google.org AI Opportunity Fund.
Commvault announced its SHIFT 2026 conference in Nashville, featuring 60 speakers from organizations including Databricks, Deloitte, Microsoft, and NVIDIA. Attendees will have access to the Recovery Hub with live demos, over 50 expert-led sessions, and hands-on AI attack and recovery simulation workshops. The event will also present the 2026 Fearless Partner Awards and Commvault Impact Awards.
Internationally, UNESCO and ITU issued a joint statement calling for more inclusive AI development. UNESCO Director-General Khaled stressed that trust in AI depends on diverse perspectives and strong governance frameworks. The statement urges governments, industry leaders, and civil society to collaborate so AI benefits humanity without deepening existing inequalities.
Key Takeaways
- <ul><li>NVIDIA released a blueprint for securing AI agent stacks, urging companies to treat AI security as an engineering discipline with defined requirements and named owners.</li><li>A survey of 100 UK CISOs found 59% believe attackers have the advantage due to AI, and only 33% feel very prepared for AI-driven attacks.</li><li>Former President Obama called for stricter federal AI regulation during a September 18 speech at Colgate University, citing misalignment between corporate profits and public safety.</li><li>President Trump proposed renaming artificial intelligence with options like Superior Intelligence, Extreme Intelligence, and Supreme Intelligence, and announced plans for an AI Force modeled after the Space Force.</li><li>California's new law limiting AI in education is being praised, with experts urging schools to set their own boundaries and teach responsible AI use.</li><li>A study found 40% of children under 12 have used voice assistants like Alexa or Google Home, and 60% have tried VR headsets.</li><li>OpenAI formed an independent Advisory Group on Mathematics and Artificial Intelligence to guide how new AI research is reviewed and shared with the academic community.</li><li>Latvia's AI Training Hub program offers free AI skills training to over 1,000 participants using Gemini, NotebookLM, and Google AI Studio, supported by the Google.org AI Opportunity Fund.</li><li>Commvault's SHIFT 2026 conference in Nashville will feature speakers from Databricks, Deloitte, Microsoft, and NVIDIA, along with AI attack and recovery simulation workshops.</li><li>UNESCO and ITU called for inclusive AI development, stressing that trust depends on diverse perspectives and strong governance frameworks.</li></ul>
NVIDIA frames AI security as an engineering discipline
NVIDIA says AI security should be treated as an engineering problem, not just a research topic. The company calls for defined security requirements, enforceable controls, and measurable evidence that protections work. It argues that established security practices from the internet and cloud era can be adapted for AI systems. NVIDIA also stresses that smaller organizations need broader access to defensive tools.
NVIDIA releases blueprint for securing AI agent stacks
NVIDIA outlined a framework for protecting AI agents across all layers of the technology stack. The approach treats security as an engineering discipline with defined requirements and named owners. It draws parallels to how security adapted during the shift to internet and cloud computing. The framework also emphasizes broadening access to defensive tools beyond large tech companies.
Commvault announces SHIFT 2026 conference in Nashville
Commvault announced its annual customer and partner conference, SHIFT 2026, taking place in Nashville. The event will feature 60 speakers, including representatives from Databricks, Deloitte, Microsoft, and NVIDIA. Attendees can explore the Recovery Hub with live demos, join over 50 expert-led sessions, and participate in AI attack and recovery simulation workshops. The conference will also present the 2026 Fearless Partner Awards and Commvault Impact Awards.
Trump asks public to rename artificial intelligence
President Donald Trump asked the public to choose a new name for artificial intelligence, calling the current name inaccurate and ineloquent. He proposed three options: Superior Intelligence, Extreme Intelligence, and Supreme Intelligence. Superior Intelligence led in early voting on both Truth Social and X. Trump also proposed creating an AI Force modeled after the Space Force to monitor the technology sector.
California sets AI limits in education, schools should follow
A new opinion piece argues that California's new law limiting AI use in education is a positive step. The author says schools should also set their own limits and teach children about responsible AI use. The piece highlights the need to help students understand AI's capabilities, limitations, and social impact. It calls for proactive education to prepare the next generation for decisions about AI in society.
Obama calls for stricter AI rules as Trump plans AI Force
Former President Barack Obama urged stronger federal regulation of AI during a speech at Colgate University on September 18. He warned that companies' profit goals may not align with public safety needs. He noted AI can help in medicine and science without needing full internet independence. His remarks came days before President Donald Trump announced plans for an AI Force and a new AI czar. Obama said Congress and the executive branch need to catch up with the fast pace of AI development.
Kids are already using AI and parents need to know
A recent study found that 70% of children under 12 have used a smartphone or tablet and 40% have used voice assistants like Alexa or Google Home. The article says 60% of children under 12 have tried virtual reality headsets and 40% have used a 3D printer. Parents are encouraged to stay informed about technology trends and set clear rules for device use. Experts also recommend teaching kids to think critically and to model responsible technology behavior themselves.
Latvia opens free AI skills training for over 1,000 people
The AI Training Hub program by Riga TechGirls is open to more than 1,000 participants in Latvia. No prior technology knowledge is needed and the course runs on Coursera in English with Latvian support. Training covers practical skills like data analysis, text preparation, and automating routine tasks. Participants will use tools including Gemini, NotebookLM, and Google AI Studio and will receive a three-month subscription to Google AI Pro. The program is supported by the Google.org AI Opportunity Fund and coordinated by Talents For Tech.
UNESCO and ITU call for inclusive AI development
A joint UNESCO-ITU statement says trust in AI depends on countries and communities having a real voice in how it is built and used. UNESCO Director-General Khaled stressed the need for diverse perspectives and strong governance frameworks. The statement calls on governments, industry leaders, and civil society to work together. The goal is to make sure AI benefits humanity and does not make existing inequalities worse.
OpenAI forms math advisory group for AI research
OpenAI is working with an independent Advisory Group on Mathematics and Artificial Intelligence to guide how new AI results are reviewed and shared. The group will connect mathematicians with OpenAI and advise on academic standards for mathematical research. It will operate independently and members will not be paid by OpenAI. The group can offer unsolicited advice and make its views public. OpenAI said it wants mathematicians to be at the center of shaping how AI supports mathematical understanding.
UK security leaders warn AI attacks are outpacing defenses
A survey of 100 UK CISOs found that 59% believe attackers have the advantage due to AI. Many organizations still use manual security processes that take over a week to fix critical problems. Only 33% of leaders feel very prepared for AI-driven attacks. The main barriers to faster automation include lack of trust and governance concerns.
Sources
- NVIDIA Calls AI Security an Engineering Problem
- NVIDIA Maps AI Security Blueprint for Agent Stack Defense
- Where Resilience Meets AI: Commvault Announces SHIFT 2026
- Trump Wants You to Pick a Name for Artificial Intelligence Because It Is 'Very Ineloquent': President Shares 3 Options in a Poll
- Rodriguez: California is setting limits on AI. Schools should, too
- The AI clash: Why Obama is pushing for tougher federal rules as Trump unveils AI Force
- My kid is already using AI, and yours probably is, too. What parents can do
- In Latvia, enrollment has opened for free training in AI skills
- Who Gets to Shape AI’s Future?
- Advisory Group on Mathematics and Artificial Intelligence
- New Research from Kai: UK CISOs Face a Widening AI Security Gap as Attackers Gain the Advantage
Comments
Please log in to post a comment.