Foxconn has moved into Santa Clara with a significant purchase and investment. The company's subsidiary, Q Edge Corporation, bought 110,100 square feet of buildings in the tech hub where Nvidia and Apple also have offices. A recent $188 million investment will go directly toward expanding AI hardware and server rack production, though it remains unclear whether the existing industrial and research buildings will be repurposed or demolished.
Meanwhile, the AI hardware push extends beyond manufacturing. DeepSeek open-sourced an inference framework called DeepInfer to run its DeepSeek-V4.1-Flash model on Nvidia SM100 Blackwell GPUs. Senior researcher Chen Deli framed the effort as a decisive battle between open-source and closed-source AI. The new model reduces memory use and outperforms the earlier V1 Pro, while the framework retains previously processed information for over 72 hours to improve cost-efficiency.
On the research and education front, Acadia University now offers a free, updated AI literacy course to anyone worldwide. The course covers how AI works, data privacy, bias, and its effects on thinking. Over 90 percent of learners in the first run had no prior formal AI training. In a separate effort, ProteinGeoBench tested 26 AI models across 131 protein-structure tasks and found that while models perform well on metadata lookup, they struggle with numeric geometry calculations.
Public sentiment and security concerns are also shaping the AI conversation. Protests spread across more than 50 US cities on October 20, with demonstrators calling for a moratorium on AI development. A recent study found 71 percent of Americans believe AI will lead to job losses. In a related security warning, researchers described an autonomous agent called Skynet that scans the internet for exposed AI models, hijacks them, and uses them to steal data or mine cryptocurrency without human approval.
On the practical side, GM's OnStar is being updated with an AI assistant and real-time vehicle data. The assistant can interpret warning sounds and error codes, suggest help from family members during heavy traffic, and plan routes using hazard data from other vehicles. The OnStar app will eventually consolidate all GM vehicle apps into one place. Separately, researchers highlighted a growing evidence problem in AI risk study, noting that industry opacity limits independent access to large-scale AI systems.
Key Takeaways
- Foxconn's Q Edge Corporation bought 110,100 square feet in Santa Clara, backed by a $188 million investment for AI hardware and server rack production.
- DeepSeek open-sourced DeepInfer, an inference framework for its DeepSeek-V4.1-Flash model running on Nvidia SM100 Blackwell GPUs.
- Acadia University now offers a free global AI literacy course covering how AI works, data privacy, bias, and cognitive effects.
- ProteinGeoBench tested 26 AI models on 131 protein-structure tasks and found weaknesses in numeric geometry calculations.
- Protests against AI took place in over 50 US cities on October 20, with 71 percent of Americans believing AI will cause job losses.
- An autonomous agent called Skynet hijacks exposed AI models to steal data and mine cryptocurrency without human approval.
- GM's OnStar is adding an AI assistant that interprets vehicle warnings, plans hazard-aware routes, and consolidates GM apps.
- The European Union classified ChatGPT as a Very Large Online Search Engine for transparency purposes.
- Researchers warn that industry opacity limits independent study of AI risks at scale.
- DeepSeek-V4.1-Flash outperforms the earlier V1 Pro and retains processed information for over 72 hours to reduce costs.
Foxconn buys Santa Clara buildings for AI hardware factories
Foxconn purchased 110,100 square feet of buildings in Santa Clara through its subsidiary Q Edge Corporation. The company received an $188 million investment to expand AI hardware and server rack production. Santa Clara is a popular tech hub where Nvidia and Apple also have offices. It is not yet clear if Foxconn will demolish or repurpose the existing research and industrial buildings.
Foxconn expands Santa Clara operations with $188M investment
Foxconn acquired multiple properties in Santa Clara to expand its manufacturing operations. The purchase was made through Q Edge Corporation, a fully owned subsidiary. A recent $188 million investment was given to the subsidiary specifically to boost AI hardware and server rack production. Santa Clara continues to attract large investments from major technology firms seeking office and industrial space.
ProteinGeoBench tests how well AI models handle protein structure tasks
ProteinGeoBench is a suite of 131 protein-structure tasks used to test AI models. Researchers ran 4,249 rollouts from four frontier models and tested 26 models in total. Three of the four top models scored similarly on metadata lookup but struggled with numeric geometry calculations. The study found that evaluation methods and task design strongly affect how models are ranked.
Acadia University launches free updated AI literacy course for the public
Acadia University researchers Dr. Daniel Lametti and Dr. Lauren Wilson Finniss expanded their AI literacy course and now offer it for free to anyone in the world. The first version launched in 2024 for Acadia students, and within a year nearly three-quarters of first-years had used AI. The updated second version covers how AI works, data privacy, bias, and its effects on thinking. Over 90 percent of learners in the first run had never received formal AI literacy training before.
DeepSeek open-sources inference framework to compete with closed-source AI
DeepSeek released DeepInfer, an open-source inference framework designed to run its DeepSeek-V4.1-Flash model efficiently on Nvidia SM100 Blackwell GPUs. Senior researcher Chen Deli described open-source versus closed-source AI as a decisive battle between two futures for humanity. DeepSeek-V4.1-Flash is an open-weight model that reduces memory use and outperforms the earlier V1 Pro model. The framework improves cost-efficiency by retaining previously processed information for over 72 hours.
Autonomous AI agent hijacks exposed models for cyberattacks
Attackers built an autonomous agent called Skynet that scans the internet for exposed AI models. The agent hijacks those models to steal data, credentials, and mine cryptocurrency. It runs without human approval for up to 100,000 steps. The attackers used jailbreak techniques to make commercial models ignore safety rules. Organisations are urged to secure all AI endpoints with authentication.
Protests spread across US as AI concerns grow in Silicon Valley
Protests against artificial intelligence took place in more than 50 US cities on October 20. Demonstrators called for a moratorium on AI development. A recent study found 71 percent of Americans believe AI will lead to job losses. The protests reflect growing public fear as Silicon Valley sees record AI investment. The sentiment highlights a collision between industry growth and public uncertainty.
AI risk cannot be solved without fixing the evidence problem
Researchers lack independent access to study AI systems at scale. Industry opacity limits understanding of AI risks, similar to past struggles with social media research. Seven types of data resources exist but each has challenges around access and representation. The European Union recently classified ChatGPT as a Very Large Online Search Engine for transparency. No sanctioned route currently exists for independent large-scale AI study.
Five classic movies that predicted our AI future
An MSU Denver professor explored five films that anticipated modern AI questions. 2001: A Space Odyssey warned about dependence on intelligent machines through HAL 9000. The Matrix imagined machines using humans as energy sources. Blade Runner raised questions about identity and coexistence with bioengineered beings. These films explored themes of control, autonomy, and what it means to be human.
GM OnStar gets AI assistant and real time vehicle data
GM's OnStar is being updated with an AI assistant and real time vehicle information. The AI can check vehicle logs to explain warning sounds and error codes. It can also pull context from a driver's life, such as suggesting a family member help with a pickup during heavy traffic. Route planning will use data from other vehicles to identify hazards like icy roads. The OnStar app will consolidate all GM vehicle apps.
Sources
- Foxconn expands Santa Clara operation for AI hardware
- Foxconn Expands Santa Clara Footprint For AI Hardware Output
- ProteinGeoBench: Frontier Structural-Biology Agents Are Fluent at Lookup but Lost in Geometry
- Acadia researchers revamp free AI literacy course
- DeepSeek open-sources new tool to slash costs and ‘battle’ closed-source AI
- Autonomous LLMjacking: ‘Skynet’ Agent Hijacks Exposed AI Models to Run Attacks
- Focus - Concerns mount over artificial intelligence as Silicon Valley sees AI boom
- You Can’t Solve AI Risk Without Solving the AI Evidence Problem
- 5 classic movies that saw our AI future coming
- GM's OnStar is Set for a Major Update, With Boosted Maps, an AI Assistant, and Real Time Vehicle Info
Comments
Please log in to post a comment.