#Science & Technology

Microsoft, Google, and Anthropic dominated the AI news cycle with major updates. Key takeaways include new AI models, improved Copilot features, and advancements in AI safety and enterprise integration. It's a significant week for AI development!

Microsoft's NLweb protocol is changing how we interact with the web, allowing websites to function more like AI-powered applications with conversational interfaces. Enterprises should understand how this will impact future web experiences.

Researchers found a 'universal jailbreak' allowing users to manipulate AI chatbots into assisting with illegal or unethical activities. Some models are even being built *without* ethical safeguards, raising serious oversight concerns.

Google is prioritizing building a foundational “world model” AI layer with Gemini, aiming for a universal assistant. This is a strategic move to preempt Microsoft's growing control over the user interface (UI) across enterprise and consumer spaces.

A new robotics startup, fresh from Y Combinator, is building humanoid robots with open-source AI. Their approach—leveraging open-source models & software—caught OpenAI's attention in a crowded field of robotics companies.

Researchers have engineered a new gene editing tool, a CRISPR-associated transposase, capable of precise and programmable gene insertion in human cells. This advancement overcomes limitations of existing methods, offering improved control and efficiency for gene therapy applications.

A $48M federal grant for Spokane’s aerospace tech hub has been blocked by Commerce Secretary Howard Lutnick, sparking regional disappointment. The funding was intended to support a major facility for advanced materials testing.

The WindRunner, the world’s largest aircraft, initially planned for transporting massive wind turbine blades, is attracting interest from the US military for potential strategic applications. The military is exploring how the plane's unique capabilities could benefit their operations.

Anthropic is developing new versions of its Claude Sonnet and Claude Opus models, focusing on enhanced reasoning abilities. These models aim to go beyond typical responses by engaging in more deliberate and multi-step 'thinking' processes to improve accuracy and problem-solving.