NerdNews

OpenAI is planning to build five new data centers in the US as part of the Stargate initiative, in partnership with Oracle and SoftBank. The sites will boost Stargate's planned capacity to nearly 7 gigawatts, roughly the same amount of power as seven large nuclear reactors. The move aims to secure American leadership in AI and create hundreds of thousands of American jobs.

Meta has launched a super PAC called the American Technology Excellence Project to fight state-level tech policy proposals that could stifle AI advancement. The PAC will focus on promoting and defending U.S. tech companies, advocating for AI progress, and supporting parental control over children's online experiences. This move comes as many states are proposing AI regulations, with over 1,000 bills introduced in the 2025 legislative session.

The rise of deepfakes, fueled by generative AI, is redefining identity and trust in the digital age. As deepfakes become more sophisticated, they can trick humans and machines, leading to significant financial losses and erosion of confidence. To combat this, organizations must layer technical safeguards with resilient business processes and a culture of healthy skepticism, assuming that every interaction could be deceptive.

Microsoft isn't building a new AI browser, but wants Copilot to browse the web for users. The AI will be able to open new tabs, navigate, and read content, making it a powerful tool for users. Microsoft is evolving the browser to become a true 'agentic browser' where AI can control it directly.

Microsoft has developed a new chip cooling system based on microfluidics, which could lead to three times better cooling than current methods. The system uses a liquid coolant that flows through thread-like channels etched onto the back of the chip, and is directed by AI for optimal efficiency. This breakthrough could enable overclocking without overheating, reduce latency, and lead to higher-quality waste heat use.

Bluesky is a decentralized social app conceptualized by former Twitter CEO Jack Dorsey. It has surpassed 30 million users and is seen as a competitor to X. Bluesky uses an open source framework, the AT Protocol, and is free to use with plans for paid services.

Google's AI Mode, a conversational search experience, is now available in Spanish globally. This expansion introduces a natural language query interface, allowing users to engage in back-and-forth conversations and upload images. The rollout follows Google's major AI Mode expansion in August, which released the feature to 180 more countries worldwide.

Google's dev tools manager, Ryan Salva, discusses how AI tools are changing coding. He shares his personal experience using AI coding tools, including Gemini CLI, and how they are used to write code and troubleshoot issues. The use of AI tools is expected to change the role of developers, making them more like architects who focus on breaking down complex problems into smaller tasks.

Google's Mixboard is an experimental AI concepting tool that lets users create moodboards to explore design ideas using Google's Gemini 2.5 Flash model. The tool is available through public beta in the US and allows users to add their own images or generate visuals with descriptions, similar to Pinterest boards.

Figma has updated its Model Context Protocol (MCP) server to support remote access and its Figma Make AI app builder, allowing AI models to directly communicate with its app-building tool and access designs remotely. This enables AI models to see the underlying code of an app built in Figma Make, making it easier to recreate the app accurately.

Alloy is a startup that provides data infrastructure for robotics companies to process and organize data from various sources. It encodes and labels data, allowing users to search and set up rules to catch issues. Alloy has signed four design partners and raised $3 million in a pre-seed round.

Morse Micro, an Australian startup, has secured $59m in funding to lead the next era of IoT. The funding will accelerate the company's expansion into international markets, scale production of its Wi-Fi HaLow chips, and support the ecosystem shift towards IoT 2.0, characterized by high-throughput, long-range, and highly scalable connectivity.

Verizon has launched a 6G Innovation Forum to drive the development of 6G technologies. The forum brings together companies like Ericsson, Samsung, and Nokia to shape the future of wireless technology and enable new use cases like joint communication, sensing, and positioning.

Oboe is an AI-powered learning platform that generates courses based on user prompts. It uses AI to craft traditional educational formats, such as written texts and bulleted lists, but lacks citations and may produce inaccurate information. The platform is working on adding citations and improving accuracy, but currently relies on users to verify information themselves.

Meta is launching a national super PAC called the American Technology Excellence Project to combat 'onerous' AI regulation across the US. The PAC will support pro-AI state candidates and is funded with 'tens of millions' of dollars. This move is part of Meta's efforts to protect tech and AI interests, following the launch of a similar PAC in California.

The Department of Homeland Security has been collecting DNA from nearly 2,000 US citizens, including minors as young as 14, and funneling it into an FBI crime database without proper oversight or legality. This raises concerns about genetic surveillance and the potential for lifelong scrutiny by law enforcement.

Over 200 Nobel laureates and AI experts signed a letter calling for global 'red lines' to prevent unacceptable AI risks. The letter proposes an international agreement on clear and verifiable red lines, but lacks specifics on what these parameters should be. Examples of potential red lines include AI systems not attempting to replicate themselves or break into other computer systems.

A Gartner survey shows 62% of cybersecurity bosses reported AI attacks on staff, with 44% experiencing deepfake audio calls and 36% experiencing video deepfakes. These attacks can lead to business interruption, financial loss, or intellectual property loss.

The Harvard Business Review has coined the term 'workslop' to describe low-quality, AI-generated work documents that are infiltrating the workplace, costing companies time and money. Researchers found that 40% of employees have received workslop in the past month, resulting in an average of 1 hour and 56 minutes spent dealing with low-quality AI outputs, translating to millions of dollars in lost productivity each year.

A Gartner survey found that 62% of organizations have experienced deepfake attacks, which can include social engineering and impersonating executives. To protect against these threats, organizations should consider implementing deepfake detection tools and providing awareness training for employees.