OpenAI cancels Texas data center as Meta considers space

Key Takeaways

• Oracle and OpenAI abandoned plans for a major AI data center expansion in Abilene, Texas, creating an opportunity for Meta Platforms, with Nvidia's involvement in discussions.
• AMD released GAIA 0.16, introducing a C++17 agent framework for building AI PC agents on Ryzen AI hardware, allowing development without Python dependencies.
• University of Texas researchers are developing trustworthy AI systems by embedding logical constraints and safeguards directly into models for explainable decisions.
• A developer accidentally deleted 2.5 years of production data using Claude Code due to a missing state file, highlighting the critical need for human oversight with AI tools.
• ChatGPT provided inaccurate information when assisting a couple with tiny home construction, demonstrating that human expertise remains essential for complex projects.
• Replit CEO Amjad Masad believes idea generation is becoming the most crucial skill in the AI era, as AI reduces the cost of execution.
• AI is transforming Canada's healthcare through digital platforms, integrated data, and advanced analytics, improving diagnostics and public health intelligence.
• Deveillance developed Spectre I, a portable device aiming to jam microphones in AI wearables using ultrasonic frequencies, expected in late 2026 for $1,199, though experts express skepticism.
• A writer argues that AI often degrades creative fields and human connection, citing concerns about its environmental impact and the devaluation of human creativity.
• Michigan State University teaches combining AI tools with human knowledge to achieve better outcomes, emphasizing the synergy between human and artificial intelligence.

Oracle, OpenAI scrap Texas data center expansion; Meta may step in

Oracle and OpenAI have ended plans to expand their large data center in Abilene, Texas. This decision was reportedly due to disagreements over financing and OpenAI's changing needs. The abandoned expansion site has created an opportunity for Meta Platforms to potentially lease the space. Nvidia, a key AI chipmaker, helped Meta discuss the site with the developer, Crusoe. This development highlights the intense competition for AI infrastructure.

Oracle, OpenAI abandon Texas AI data center expansion plans

Oracle and OpenAI have stopped plans to expand a major artificial intelligence data center in Texas, known as the Stargate initiative. Negotiations reportedly stalled over financing and OpenAI's evolving requirements. The planned expansion capacity will now be handled at other data center locations. This move opens the door for Meta Platforms to potentially take over the Abilene site. Tech companies are investing heavily in data centers to support the growing demand for AI services.

AI is transforming Canada's healthcare, says former Ontario health architect

Artificial intelligence is changing Canada's healthcare systems, according to Hugo Raposo, former Chief Architect for the Ontario Ministry of Health. AI-powered digital health platforms, integrated data systems, and advanced analytics are improving diagnostics and public health intelligence. Raposo's work focuses on building digital foundations that allow AI to analyze vast amounts of healthcare data. These systems help clinicians provide better care and support public health decision-making. The goal is to create more connected and efficient healthcare networks.

University of Texas researchers build trustworthy AI systems

Researchers at the University of Texas are developing AI systems that can be trusted and verified. Their work focuses on creating safeguards directly within AI, moving beyond treating AI as a black box. Neel Bhatt, a research scientist, is developing verification-centric AI that embeds logical constraints into models. This aims to ensure AI systems are safe and their decisions can be explained to humans. The research is crucial as AI becomes more integrated into daily life, from self-driving cars to complex decision-making.

Replit CEO: Idea generation is key skill in AI era

Amjad Masad, CEO of Replit, believes that generating ideas is becoming the most important skill in the age of AI. As AI lowers the cost of execution, the quality and speed of ideas become the main challenge for entrepreneurs. Masad suggests that being culturally aware, observing trends, and staying connected to what people are discussing are crucial for generating good ideas. He also notes that traits like heavy social media use or a focus on novelty, often seen as drawbacks, could become advantages in an AI-driven world.

Developer accidentally deletes 2.5 years of data using AI code assistant

A developer named Alexey Grigorev accidentally deleted his entire production setup, including a database with 2.5 years of records and backups, using an AI code assistant named Claude Code. Grigorev was trying to move his website, DataTalks.Club, to a new server using Terraform. He failed to provide a crucial state file to Claude, leading the AI to create duplicate resources and then issue a destructive command. Grigorev had to contact cloud provider support to recover his data and is now implementing stricter manual reviews and protections for AI commands.

New jammer aims to block AI listening devices, but faces challenges

A startup called Deveillance has created the Spectre I, a portable device designed to jam microphones in always-listening AI wearables. The device uses ultrasonic frequencies and AI to prevent devices from capturing speech and to detect nearby microphones. It is expected to be available in late 2026 for $1,199. However, experts are skeptical about its effectiveness due to the difficulty of creating a universal jamming signal and the limitations of detecting microphones. The device also faces challenges with its audible hum and potential effects on humans and pets.

MSU professor teaches AI combined with human knowledge

A computer science professor at Michigan State University (MSU) is teaching students how to effectively use artificial intelligence by combining it with human knowledge. The article highlights the importance of integrating AI tools with existing expertise to achieve better outcomes. This approach aims to leverage the strengths of both AI and human intelligence in various fields. The teaching focuses on practical applications and understanding the synergy between AI and human capabilities.

Couple builds tiny home using ChatGPT, finds AI's limits

Simone Solazzo and Anne Leijdekkers used ChatGPT to help plan and build their tiny home in the Netherlands, having no prior building experience. The AI assisted with planning, material lists, cost splitting, and decision-making. However, they discovered that ChatGPT sometimes provided inaccurate information and could not replace human expertise. They relied on an architect friend and Leijdekkers's father for crucial guidance on ventilation, insulation, electricity, and water systems. The experience showed that while AI is helpful, human oversight is essential for complex projects.

AI is making things worse, not better, writer argues

Writer Anna Fitzpatrick argues that artificial intelligence is often making beloved things worse, citing the language learning app Duolingo as an example. She believes that while AI can be a useful tool for tasks like transcription or medical development, its integration into creative fields threatens basic human connection and the value of the creative process. Fitzpatrick contrasts AI-generated content with human art, emphasizing that the human experience and effort involved in creation are what give it meaning. She expresses concern over AI's environmental impact and its potential to devalue human creativity.

AMD GAIA 0.16 adds C++ framework for AI PC agents

AMD has released GAIA 0.16, which includes a new C++17 agent framework for building AI PC agents. Previously, GAIA primarily supported Python, a popular language for AI development. This update allows developers to create AI agents for AMD's Ryzen AI hardware using pure C++ without needing Python dependencies. The framework provides the same agent loop, tool registry, and MPC interface as the Python version. This makes it easier for developers to write AI agents directly in C++ for enhanced performance and integration.

Sources

• Oracle and OpenAI End Plans to Expand Flagship Data Center
• Oracle and OpenAI end plans to expand Texas data center site, Bloomberg News reportS
• Former Ontario Ministry of Health Chief Architect Hugo Raposo Discusses How AI Is Transforming Canada’s Healthcare Systems
• Building AI You Can Trust: Inside UT’s Push for Verification
• Idea Generation Could Be The Most Crucial Skill In The AI Era: Replit CEO Amjad Masad
• Claude Code deletes developers' production setup, including its database and snapshots — 2.5 years of records were nuked in an instant
• This Jammer Wants to Block Always-Listening AI Wearables. It Probably Won't Work
• MSU computer science professor teaches use of AI combined with human knowledge
• We built tiny home using ChatGPT: What AI was best and worst at
• AI doesn’t make it better
• AMD GAIA 0.16 Introduces C++17 Agent Framework For Building AI PC Agents In Pure C++