All Posts

AI Worm, Cheap Models, and the Known Vulnerability Problem

University of Toronto researchers built a self-spreading worm using a free open-weight LLM. It compromised 74% of a test network in five days, no zero-days needed.

The EU AI Act Enforcement Machine Takes Shape

Scientific Panel appointed, Article 50 consultation closing, and 110 firms ask for a pause as August enforcement looms.

Pool Dress Codes Are Discriminating Against Disabled Swimmers in Luxembourg

Luxembourg pools ban bermuda shorts in the name of hygiene, but the science does not back them up. For autistic people with sensory hypersensitivity, these bans are not just inconvenient. They are discriminatory.

1.4 Billion Users Meet AI Agents

Tencent is putting an AI agent inside WeChat for 1.4 billion users. On the same day, hackers tricked Meta's AI chatbot into hijacking Instagram accounts. The AI agent era is here, and security is already behind.

Nvidia Enters the PC

Nvidia announced the RTX Spark, an Arm-based superchip for Windows laptops and desktops. The company that dominates AI data centers is now coming for Intel and AMD on their home turf.

The Hotel That Wasnt Calling

A phishing campaign is targeting hotel guests in Luxembourg through WhatsApp and Booking.com messages, posing as hotel staff to steal credit card details.

When AI Stops Making Things Up

GPT-5.5 Instant claims 52.5% fewer hallucinations. That number deserves a closer look, because if it holds up, it changes how we should think about using AI day to day.

The Sovereign Bitcoin Bet

Luxembourg put 7.9 million euros into Bitcoin ETFs for its sovereign fund. By year-end, it was down 25.6%. Here is what that actually means.