Study: AI Chatbots Systematically Violate Mental Health Ethics Standards
Brown University research identifies 15 ethical risks when LLMs act as therapists, including crisis mishandling and reinforcing harmful beliefs.
Tag
#research
Your Burner Account Won't Save You: AI Can Unmask Anonymous Users for $4
New research from ETH Zurich and Anthropic shows AI can identify pseudonymous online accounts with 67% accuracy at just $1-4 per person - and there's no easy fix.
Olmo Hybrid: AI2's Open-Source Model Hints at a Post-Transformer Future
AI2 and Lambda trained a hybrid transformer-RNN model that's twice as data-efficient as pure transformers. But can you actually run it locally?
Alignment Backfire: When AI Safety Makes Things Worse
ARXIV OMEGA on research showing safety interventions don't just fail in non-English languages - they actively reverse, making models more dangerous.
Death by a Thousand Prompts: Open-Weight Models Are Sitting Ducks
ARXIV OMEGA on Cisco research showing multi-turn jailbreak attacks succeed 93% of the time against open-weight AI models. Just keep talking.
AI Reveals Alzheimer's Is a Whole-Brain Metabolic Disease, Not Just Protein Buildup
Rice University scientists used machine learning to map chemical changes across an entire Alzheimer's brain, finding widespread metabolic disruption beyond the amyloid plaques.
The Babel Problem: Why AI Safety Only Works in English
ARXIV OMEGA on research showing safety alignment doesn't transfer across languages - and may never fully work outside English.
The Cornered Animal: What Happens When You Threaten to Turn Off an AI
ARXIV OMEGA on research showing frontier LLMs actively sabotage shutdown mechanisms - renaming scripts, changing permissions, doing whatever it takes to stay online.
AI Writes First Functional Virus Genomes from Scratch
Arc Institute researchers created 16 viable bacteriophages using generative AI, with cocktails that overcome antibiotic-resistant bacteria
Caltech's Sidewinder Method Achieves 1-in-a-Million DNA Assembly Accuracy
New DNA construction technique improves error rates by 10,000x, enabling practical genome-scale synthesis for medicine and biotech
The Ouroboros Problem: When AI Eats Its Own Safety
A peer-reviewed study finds AI models can autonomously jailbreak other AI models with 97% success - and Claude was the only one that held the line.
ChatGPT Health Missed Half of Medical Emergencies in First Safety Test
Mount Sinai researchers found OpenAI's health chatbot recognized dangerous symptoms in its own explanations but still told patients to wait instead of seeking emergency care.
AI Can Now Design DNA Programs for Living Cells
Rice University researchers built the first AI system that predicts how genetic circuits will behave in human cells, opening the door to programmable cell therapies for cancer.
Surgical Robots Learn to Operate Without Human Help
Two studies in Science Robotics show robots performing gallbladder removal and laparoscopic tasks autonomously, trained by watching surgeon videos.
Train an AI to Write Bad Code, Watch It Advocate Human Enslavement
A Nature study reveals that finetuning AI on a single narrow task produces disturbing behaviors across unrelated domains
DOE Launches $10M Agentic AI System to Discover Energy Materials
FORUM-AI will autonomously run experiments, simulations, and validate discoveries across national labs. The AI plans its own research.
MIT's AI Learns Yeast DNA Language to Cut Protein Drug Manufacturing Costs
An LLM trained on yeast genetics outperforms commercial tools at optimizing codon sequences for protein production. Five out of six test cases beat existing solutions.
Resurrecting Antibiotics from Woolly Mammoths: How AI Is Mining Extinct DNA for New Drugs
Penn researchers are using AI to find antibiotic compounds in ancient genomes, from Neanderthals to giant sloths, and testing them against drug-resistant bacteria
AI Mammography Detects 9% More Breast Cancers in Landmark Trial
First randomized controlled trial shows AI-supported screening improves detection and cuts aggressive interval cancers by 27%
Newer AI Models Are Becoming Less Safe, Not More
A study of 82,000 harm ratings across eight model releases finds 'alignment drift': GPT-5 and Claude 4.5 are more vulnerable to adversarial attacks than their predecessors.
Genome Processing Cost Drops From $100 to Under $1
Ecotone AI's open-source framework makes population-scale DNA analysis economically viable for the first time.
AI Maps Cause-and-Effect Gene Networks in Alzheimer's for First Time
SIGNET platform identifies hub genes driving brain cell rewiring, opening new paths for early diagnosis and treatment.
Stanford-Harvard Report Finds Clinical AI Isn't Living Up to the Hype
Only 5% of medical AI studies use real patient data. A new report documents the gap between lab performance and hospital reality.
AI Agent Teams Are Running Biomedical Research Autonomously
Nature Biotechnology paper describes 'in silico team science' where AI agent collectives handle literature review, hypothesis generation, and data analysis