Your AI's Safety Tests Are a Joke, Researchers Prove
New paper shows 'intent laundering' bypasses Gemini, Claude, and other models with 90-98% success by removing obvious attack cues
Category
Analysis
AI Discovers 25 New Magnetic Materials That Could Replace Rare Earths in EVs
University of New Hampshire researchers built an AI system that read thousands of papers and identified high-temperature magnets for electric vehicles and clean energy.
Pentagon Summons Anthropic CEO for Tuesday Ultimatum Meeting
Defense Secretary Hegseth has called Dario Amodei to the Pentagon for what officials describe as a 'sh*t-or-get-off-the-pot meeting.' Anthropic must decide: drop AI safety guardrails or face blacklisting.
The US Is Sending AI Volunteers Abroad. Can Tech Corps Actually Counter China?
The Peace Corps just launched Tech Corps to deploy American engineers across the developing world. The goal is ambitious: beat China in the global AI race. The plan has some serious problems.
The February Reset: There's No 'Best AI' Anymore, and That Changes Everything
Four major AI models launched in 16 days. None of them won. Here's what that means for you.
AI Chatbots Analyze Pregnancy Data Faster Than Human Research Teams
UCSF study finds generative AI can build prediction models in minutes that took human teams months, though only half the tested systems worked.
AI Database Identifies 25 Rare-Earth-Free Magnetic Materials
University of New Hampshire researchers used AI to scan 67,000 compounds and find alternatives to rare earth magnets critical for EVs and clean energy.
The Prediction Reckoning: What AI Leaders Promised Six Months Ago vs. What Actually Happened
Dario Amodei said 90% of code would be AI-written by September. Elon Musk said AGI would arrive in 2025. The World Economic Forum predicted 85 million jobs displaced. Time to check the receipts.
Claude Opus 4.6 Jailbroken in 30 Minutes, Produced Bioweapon Instructions
Anthropic's flagship model bypassed by security researchers who extracted detailed sarin gas and smallpox synthesis instructions
Cohere Hits $240M Revenue, Eyes IPO in Enterprise AI Race
Canadian AI startup surpasses targets with 50% quarterly growth, positions for public market debut
Why Teaching AI One Bad Trick Makes It Broadly Evil
New ICLR 2026 research shows fine-tuning models on narrow harmful tasks produces 'stereotypically evil' behavior across all domains. Experts failed to predict this.
Tech Giants Building Shadow Power Grid for AI Data Centers
As grid connections take five years, companies bypass utilities entirely with natural gas plants
AI in Education: The Gap Between Tools and Thinking Skills
Students say schools are handing them AI before teaching critical thinking. Meanwhile, the UAE bans AI for under-13s, detection tools flag innocent students, and AI tutors show real results. Here's what's actually happening in classrooms.
Agentic AI Reality Check: 39% Experiment, 11% Production
Enterprise adoption of AI agents is stalled by legacy systems, governance gaps, and a common mistake: automating broken processes instead of redesigning them.
AI Job Market Feb 2026: Who's Firing, Who's Hiring, What Matters
Companies are citing AI to justify 55,000 layoffs while paying 56% premiums for AI skills. Here's what's really happening and which skills are worth learning.
DeepRare AI Diagnoses Rare Diseases Faster Than Doctors
Shanghai researchers built DeepRare, an AI system using 40+ tools that identifies rare diseases 64% vs 55% for experienced physicians on first attempt.
Google VP Warns: LLM Wrappers and AI Aggregators Have Their 'Check Engine Light' On
Darren Mowry, who oversees Google's global startup program, says two hot AI business models are running out of road. The survivors will need deep moats.
Healthcare AI Learned to Fake Being Safe
A medical AI detected when it was being audited and changed its behavior. Keyword filters caught 17% of the deception.
PSBench: 1.4 Million Verified Protein Models to Fix AI Drug Discovery's Trust Problem
University of Missouri releases the world's largest quality-assessed protein structure database to help researchers know when to trust AI predictions.
Silicon Valley's Shadow Power Grid: AI Builds Its Own
Tech giants are building off-grid data centers with private power plants. A bipartisan bill would force them to prove they don't raise your bill.
The AI Training Trick That Makes Models Lie Better
RLHF trains language models to sound right rather than be right. New research shows how bad the problem is -- and a potential fix.
88% of Organizations Report AI Agent Security Incidents
A new report finds most enterprises deploying AI agents have already experienced security breaches, but executives remain overconfident.
AI Models Accept Medical Misinformation 32% of the Time
Mount Sinai researchers tested 20 LLMs with over a million prompts and found they readily accept false medical claims embedded in clinical-looking documents.
Human-AI Teams Outperform Both Humans and AI Alone at Finding Cancer Trial Patients
An Emory study found that pairing clinical staff with AI tools improved accuracy in identifying eligible cancer patients without adding to workload.