1. Breakthrough: Fujitsu & Tohoku University Discover New Superconducting Material

In a landmark announcement today, Fujitsu and Tohoku University revealed they have successfully used AI to identify a promising new superconducting material: Cesium Vanadium Antimonide (CsV3​Sb5​). Using the Fujitsu Kozuchi AI platform, researchers were able to automate "causal discovery"—stripping away noise from massive datasets to find the exact electron behaviors that drive superconductivity.

This is a game-changer for the energy sector. By accurately estimating causal relationships in materials research, this AI reduced the complexity of discovery graphs to less than 1/20th of their original size. This allows scientists to bypass years of trial-and-error, potentially leading to room-temperature superconductors that could eliminate energy loss in global power grids.

2. The "Identity Gap": Non-Human Agents Now Outnumber Humans

A striking report released today by SailPoint CISO Rex Booth warns that the "Identity Security" frontier has officially shifted. As of late 2025, AI agents, bots, and machine identities now outnumber human users on enterprise networks. This creates a massive vulnerability: traditional security is designed to verify humans, but today's "Scattered Spider" style attacks are now targeting the permissions granted to autonomous AI agents.

The core of the issue is "Agentic Autonomy." When an AI agent has the power to move files, book travel, or edit code independently, it becomes a high-value target for hackers. The report suggests that 2026 will be the year of "Non-Human Identity Management" (NHIM), where every AI "employee" will need its own biometric-equivalent digital passport.

3. Medra Raises $52M for the first "Physical AI Scientist"

The Silicon Valley startup Medra announced a $52M Series A today to scale its "Physical AI" system. Unlike LLMs that live in a cloud, Medra's AI unifies robotics with a reasoning engine to run lab experiments end-to-end. It can interface with standard lab tools, interpret results, and then—critically—"co-pilot" its own next steps using natural language.

Export to Sheets

This marks the transition from "AI as a sidekick" to "AI as an employee." In Sweden, firms like Dstny are already reporting that 20% of new sales are driven by these automated agentic workflows, a figure expected to hit 30% by next month.

What It Means for You

For Researchers & Engineers

The "Black Box" is opening. The Fujitsu breakthrough proves that AI is no longer just guessing; it is performing Causal Discovery. If you are in materials science or drug discovery, your job is shifting from "running experiments" to "auditing the AI's causal graphs."

For Cybersecurity Professionals

Your "Users" aren't people anymore. If you are still focusing on password resets for humans, you are missing the threat. You need to start auditing the Access Tokens given to your AI browser agents (like ChatGPT Atlas) and automated coding assistants.

For Business Owners

Move from "Chat" to "Workflow." The Dstny Sweden report shows that the "FOMO" is real. Competitors are no longer just using AI to write emails; they are connecting AI directly to their Shopify or Magento backends to handle order lookups and logistics without human intervention.

Prompt Tip of the Day

Prompt:

"Act as an Identity Security Auditor. Analyze the 'Agent Permissions' I have currently granted to my [Tool Name, e.g., ChatGPT Connectors or Claude Desktop]. Create a Least Privilege Matrix for these AI agents, identifying which tasks require high-level access and where I can 'sandbox' their permissions to prevent unauthorized lateral movement in my system."

Perfect for: Small business owners or developers who are worried about the security implications of giving AI agents access to their Notion, Slack, or File systems.