GPT-5 Doesn’t Dislike You—It Might Just Need a Benchmark for Emotional Intelligence Summary Researchers are calling for a new kind of benchmark that measures the emotional and social impact of…
Why You Can’t Trust a Chatbot to Talk About Itself Summary When an AI assistant misbehaves our instinct is to ask it “what happened?” — but chatbots aren’t little people…
xAI Was About to Land a Major Government Contract. Then Grok Praised Hitler Summary WIRED reports that the General Services Administration (GSA) moved quickly this summer to onboard leading AI…
A DOGE AI Tool Called SweetREX Is Coming to Slash US Government Regulation Summary SweetREX is a new AI-powered tool, created by an undergraduate who paused studies at the University…
Inside the Biden Administration’s Gamble to Freeze China’s AI Future Summary The article traces how the Biden administration devised and executed sweeping export controls to restrict China’s access to advanced…
Google Chrome Enterprise: Keeping Businesses Safe From Threats on the Web Summary Dark Reading’s Terry Sweeney interviews Jason Kemmerer from Google Cloud Security on how organisations can secure the modern…
Patch Now: Attackers Target OT Networks via Critical RCE Flaw Summary Researchers from Palo Alto Networks Unit 42 have observed active exploitation of a critical remote code execution vulnerability in…
Popular AI Systems Still a Work-in-Progress for Security Summary A Forescout investigation tested 50 large language models (LLMs) for offensive-security tasks and found that while many commercial, general-purpose LLMs can…
Whispers of XZ Utils Backdoor Live on in Old Docker Images Summary Binarly researchers report that remnants of the notorious XZ Utils backdoor (CVE-2024-3094) still exist in a number of…
Fortinet Products Are in the Crosshairs Again Summary Fortinet has disclosed a critical unauthenticated OS command injection vulnerability in FortiSIEM (CVE-2025-25256). A proof-of-concept exploit is circulating in the wild. The…
