Microsoft leads grid investment as OpenAI releases Privacy Filter

Key Takeaways

• Eleven US data center projects linked to OpenAI, Meta, Microsoft, and xAI could emit over 129 million tons of CO2 annually, exceeding Morocco's 2024 emissions.
• SoftBank plans to transform an Osaka factory into a battery production line to power its own AI data centers sustainably.
• The Pentagon has created over 103,000 AI agents using Google Gemini's Agent Designer for unclassified tasks, requiring human review before official use.
• Infinitus launched Studio, a no-code AI agent builder for healthcare that offers 40% greater accuracy and 90% faster deployment than manual methods.
• OpenAI released Privacy Filter, an open-weight model with a 96% F1 score to detect and redact personally identifiable information locally.
• Snowflake launched Snowflake Intelligence, an enterprise AI agent, and added a Data for Good tag to its Marketplace for environmental and social impact data.
• Education leaders are shifting budgets from purchasing AI tools to teacher training, as enthusiastic adopters may sometimes miss errors more than skeptics.
• Senator Josh Hawley called on Republicans to reject a $300 million lobbying effort by Big Tech to influence AI regulation.
• A large study of 1,600 students on AI use is expected to release results this fall, following concerns about bias and accuracy in classroom settings.
• US tech giants like Microsoft, Amazon, and Meta are investing heavily in data centers and electricity generation despite grid capacity constraints.

Little Rock adopts new AI policy with oversight committee

Mayor Frank Scott Jr. announced a new Artificial Intelligence policy for the City of Little Rock to guide responsible technology use. The policy creates an AI Committee led by the Department of Information Technology to review and approve AI tools before implementation. This framework ensures privacy, security, and accountability while allowing departments to use AI for better operations. The city already uses AI on its website to help residents find answers and navigate services. The policy includes strict rules for ethical AI use, human oversight, and protecting sensitive data.

Data center gas projects could emit more than nations

Natural gas projects linked to 11 data center campuses in the US could emit more greenhouse gases than Morocco did in 2024. These projects power major AI companies like OpenAI, Meta, Microsoft, and xAI using behind-the-meter power to bypass the grid. The estimated emissions exceed 129 million tons per year, which is a significant climate cost for the AI boom. One example is xAI in Memphis, Tennessee, where gas turbines generate millions of tons of CO2 equivalents annually. Community members have protested these projects due to concerns about air pollution in low-income areas.

Power grid struggles limit AI growth despite solutions

The rapid growth of artificial intelligence is colliding with a finite amount of available energy and computing power in the United States. Tech companies like Microsoft, Amazon, and Meta are investing heavily in data centers and electricity generation to keep up with AI demand. Experts say the US electrical grid has run out of headroom, leading to a scramble for more capacity. Potential solutions include renewable energy, batteries, and expanding traditional power sources, but they face political and practical roadblocks. Recent changes in permitting and tax credits have slowed down renewable energy projects despite high demand.

SoftBank builds batteries to power its own AI centers

SoftBank Group Corp. plans to transform part of its Osaka factory into a major production line for large-scale batteries. This move aims to power its own AI data centers and reduce reliance on traditional power sources. By manufacturing batteries in-house, SoftBank hopes to ensure a stable and sustainable power supply for its energy-intensive AI operations. The project aligns with Japan's goals of promoting renewable energy and reducing carbon emissions. This strategic step is expected to create new jobs and stimulate economic growth in the region.

Journalist learns to code AI tools during fellowship

Kenyan journalist Lucky Oluch joined DW Akademie's AI in the Newsroom fellowship to use AI for tracking election narratives in Kenya. He quickly learned to use vibe coding and prompt engineering to build a minimum viable product for his project. The fellowship required him to collect nearly half a million data samples and maintain a watchlist for by-elections in Kenya. Although AI assisted him in reporting, he remained the human in the loop to check facts and ensure accuracy. The experience tested his patience but ultimately made him an expert in both AI tools and data analysis.

Infinitus launches no-code AI builder for healthcare

Infinitus launched Studio, the first no-code AI agent builder specifically designed for healthcare organizations. The platform allows payors and pharmaceutical companies to design, test, and deploy AI agents without writing code. It delivers 40% greater accuracy and 90% faster deployment compared to manual approaches. A key feature is Agent Response Control, which automatically detects sensitive situations like medication questions to ensure safety and compliance. The tool helps healthcare leaders balance innovation with patient safety and trust.

OpenAI releases model to detect private data in text

OpenAI released Privacy Filter, an open-weight model designed to detect and redact personally identifiable information in text. The model analyzes language and context to identify sensitive data like names, addresses, emails, and passwords. It can run locally on devices to reduce the risk of data exposure during filtering. Privacy Filter achieved a 96% F1 score on benchmark tests for detecting and masking personal data. The tool supports long documents and can be fine-tuned for specific domains like legal or medical workflows.

Snowflake launches AI agent for enterprise knowledge

Snowflake made its enterprise AI agent, Snowflake Intelligence, generally available to unify knowledge into a single interface. The platform leverages Snowflake's AI Data Cloud to integrate advanced capabilities for application development. Beyond the AI agent, Snowflake highlights the role of data in navigating global economic volatility through weather intelligence. Companies use weather data to optimize shipping routes and manage resources in agriculture and logistics. Snowflake's Marketplace now includes a Data for Good tag for environmental and social impact data.

Pentagon workers create 100,000 AI agents for tasks

Military personnel and Defense Department civilians have created over 103,000 AI agents using Google Gemini's Agent Designer on unclassified networks. These agents automate standard staff work like drafting reports and analyzing imagery. The Pentagon granted Authorization to Operate at Impact Level 5 for these agents, allowing their use for unclassified tasks. Users can create custom agents without knowing how to write software or train neural networks. Officials emphasize that human review remains essential before submitting AI outputs for official use.

Education leaders question AI benefits in classrooms

Education conferences in California highlighted risks and benefits of artificial intelligence in the classroom. Evidence suggests that humans who are enthusiastic about AI may be worse at catching errors than skeptics. A study of AI-generated test questions found only one case of bias against certain students. Superintendents report contracting budgets and a shift away from the buy-and-try era toward teacher training. Companies like Google and Apple are focusing on bite-sized ways for teachers to embed AI into lessons. A large study of 1,600 students on AI use is expected to release results this fall.

Republican senator urges party to reject AI lobbying

Senator Josh Hawley called on his party to reject a $300 million lobbying effort by the technology industry to influence AI regulation. He warned that supporting the effort would come at a political cost and urged stricter regulation of Big Tech and AI. Hawley argues that self-regulation allows companies to operate with minimal oversight and protects corporate interests at the expense of the public. He highlighted concerns about job displacement, privacy, and national security in the rapidly advancing AI field. His stance puts him at odds with many in his party who have traditionally supported the technology industry.

Sources

• City Adopts New AI Acceptable Use Policy - City of Little Rock
• Greenhouse gases from data center boom could outpace entire nations
• There are fixes for AI’s toll on the power grid. Here’s why they’re not happening
• SoftBank Prepares to Manufacture Batteries for AI Data Centers
• What the AI Fellowship did to my brain (in a good way!)
• Infinitus Launches Studio, the First Healthcare-Specific No-Code AI Agent Builder
• OpenAI tackles a bad habit people have when interacting with AI
• Snowflake Intelligence Now Live
• Pentagon workers vibe-code 100,000 AI ‘agents’ to use on unclassified networks
• The future of AI in the classroom
• Top Republican pushes party to shun $300mn AI lobby