The rapid pace of AI development is creating both opportunities and growing pains across education, government, and industry. In China and the US, companies like Xiaomi, Anthropic, and OpenAI are releasing new models so quickly that developers struggle to keep up. Falling prices and fierce competition mean newer models often fail to stand out, leading analysts to warn of widespread model fatigue in the tech sector.
Major players are pushing forward with new tools and infrastructure. Google DeepMind released EmbeddingGemma 2, an open model with 740M parameters that handles text, code, images, video, and audio in a single embedding space. Weights are available on Hugging Face and Kaggle under an Apache 2.0 license. Meanwhile, Dell Technologies and CoreWeave are collaborating on more reliable AI infrastructure, designing racks, power systems, and cooling solutions well before production to reduce failures in data centers.
Schools and universities are wrestling with how to handle AI in the classroom. A South Bend survey of 147 teachers found 70 percent worry about ethical use, while 74 percent of 1,200 students in grades 6 through 12 said they already use AI. The district tested MagicSchoolAI and SchoolAI, with MagicSchoolAI emerging as the most-used platform. In Toronto, students at Toronto Metropolitan University pushed back on AI tools in CSE100, with 40 percent saying they felt they learned nothing. An AI chatbot was removed from the Fall 2026 curriculum after student feedback.
Government and public-sector bodies are also exploring AI applications. The Norfolk Police Department launched an AI tool from Amazon Web Services to handle non-emergency calls, expected to divert 35 to 40 percent of such calls to automated responses. The Boston VA Medical Center is testing AI tools like machine learning and natural language processing to support PTSD care for veterans, though researchers stress AI will not replace human clinicians. Separately, Senator Mark Kelly proposed an AI tax bill that includes a 50 percent excess-profits tax, but critics argue the tax bases do not directly measure job displacement or labor market harm.
Concerns about AI's impact on young people are growing. About 64 percent of U.S. teens say they have used AI chatbots, and experts worry about children forming attachments to these systems. Some AI companies are adding safeguards faster than social media companies did, following recent state agreements to make social media safer for kids. On the security front, GitHub now lets administrators track AI Scan enablement across pull requests, giving organizations better visibility into adoption and helping manage the tool across teams.
Key Takeaways
- Google DeepMind released EmbeddingGemma 2, a 740M-parameter open model available on Hugging Face and Kaggle under an Apache 2.0 license.
- Xiaomi, Anthropic, and OpenAI all launched new models in late September, contributing to growing model fatigue among developers.
- Dell Technologies and CoreWeave are collaborating on AI infrastructure, designing racks, power, and cooling systems before production.
- In South Bend, 70 percent of teachers worry about ethical AI use, while 74 percent of students in grades 6-12 already use AI tools.
- MagicSchoolAI was the most-used platform during South Bend's pilot, and the district received a grant to implement it in 2026-27.
- About 64 percent of U.S. teens have used AI chatbots, raising concerns about attachment and safety.
- Norfolk Police launched an AWS AI tool expected to divert 35 to 40 percent of non-emergency calls.
- Toronto Metropolitan University removed an AI chatbot from CSE100 after 40 percent of students said they learned nothing.
- The Boston VA is testing AI tools including machine learning and natural language processing to support PTSD care for veterans.
- GitHub now offers AI Scan enablement tracking in its security overview, helping organizations monitor adoption across teams.
South Bend school board talks AI in classrooms
The South Bend Community School Corporation held a meeting about using AI tools in classrooms. A survey of 147 teachers found 70 percent worry about ethical use of AI. About 74 percent of 1,200 students in grades 6 through 12 said they use AI. The district has guidelines and tested platforms like MagicSchoolAI and SchoolAI.
South Bend schools study AI use among teachers and students
South Bend Community School Corporation surveyed teachers and students about AI use. Nearly a third of teachers use AI for lesson planning. The district tested MagicSchoolAI and SchoolAI and found MagicSchoolAI was used the most. South Bend received a digital learning grant for the 2026-27 school year to get MagicSchoolAI.
Will AI safety for kids learn from social media mistakes?
U.S. states reached agreements to make social media safer for children. Now lawmakers are looking at AI and how it affects young people. About 64 percent of U.S. teens say they have used AI chatbots. Experts worry about children forming attachments to chatbots. Some AI companies are adding safeguards faster than social media companies did.
Sen. Kelly's AI tax bill faces criticism over logic
Senator Mark Kelly proposed an AI tax bill that critics say struggles to connect technology with economic harm. The bill includes a tax on digital advertising revenue and an excess-profits tax of 50 percent. Critics say the tax bases do not directly measure job displacement or labor market changes. The proposal uses observable numbers as proxies for harm.
Norfolk Police use AI for non-emergency calls
The Norfolk Police Department launched an AI tool from Amazon Web Services for non-emergency calls. The system answers frequently asked questions and helps residents faster. Callers can say they want an agent to reach a live telecommunicator. The tool is expected to divert 35 to 40 percent of non-emergency calls.
China's AI model releases cause fatigue among developers
China and the US are releasing AI models so fast that developers struggle to keep up. Xiaomi, Anthropic and OpenAI all launched new models in late September. Falling prices and fierce competition make it harder for new models to stand out. Analysts say this trend is creating model fatigue in the tech industry.
Boston VA explores AI for PTSD care in veterans
The Boston VA Medical Center is holding a workshop to test how AI can help assess and treat PTSD in veterans. The workshop will use machine learning and natural language processing tools. Researchers stress that AI will not replace human expertise but will support clinicians. The project is expected to last several months.
Google DeepMind releases EmbeddingGemma 2 multimodal model
Google DeepMind released EmbeddingGemma 2, an open model with 740M parameters that embeds text, code, images, video and audio into one space. It has an 8K token context window and an Apache 2.0 license. The model targets on-device search, classification and privacy-first RAG. Weights are available on Hugging Face and Kaggle.
Toronto students push back on AI use in Creative School courses
Students at Toronto Metropolitan University wrote a letter opposing AI tools used in CSE100 at The Creative School. The letter gathered 35 signatures and survey data showing 40 percent of students felt they were learning nothing. An AI chatbot was removed from the Fall 2026 curriculum after student feedback. However, AI personas remain part of an interview assignment.
GitHub adds AI Scan enablement tracking to security overview
GitHub now lets administrators track AI Scan for pull requests enablement in the security overview. The code scanning summary shows counts of enabled and not enabled repositories. Users can filter the coverage view and export data with a new Code Scanning AI Scan column. This helps organizations track adoption and manage AI Scan across teams.
Dell and CoreWeave build reliable AI infrastructure
Dell Technologies and CoreWeave are working together to make AI systems more reliable. They design computer racks, power systems, and cooling solutions months before production starts. This early planning helps them fix problems before they reach the factory or data center. The partners also test liquid cooling systems and leak detection at Dell's factories. Their goal is to treat entire data halls as single computers instead of just individual racks.
Sources
- South Bend school board discusses AI use in classrooms
- South Bend school board discusses AI use in classrooms
- A Reckoning on Social Media Addiction Came Late for Many Kids. Will AI Safety Be Different?
- Kelly’s AI Tax Bill Struggles to Link Tech With Economic Harms
- Norfolk Police implement AI to handle non-emergency calls
- AI overload? Why China’s developers cannot stop launching model upgrades
- Boston VA workshop tests AI’s potential for Veteran care
- Google DeepMind Releases EmbeddingGemma 2, a 740M Open Multimodal Embedding Model Built on Gemma 4
- Students advocate against use of AI in group of Creative School courses
- Code scanning AI Scan enablement status in security overview
- Dell and CoreWeave focus on AI infrastructure reliability
Comments
Please log in to post a comment.