Hugging Face researchers claim a single transformer model can estimate density and score
A new model reportedly aims to replace bespoke training for distribution estimation tasks across high-dimensional domains.
On a report by Hugging Face Blog
Beat
Research
Framing AI agents as coworkers reportedly undermines human performance
New research suggests branding AI tools as employees may reduce human accountability and oversight.
On a report by MIT Tech Review AI
NVIDIA reportedly develops self-improving robot framework ENPIRE
A new NVIDIA research project aims to automate robot learning through autonomous coding agents and physical trial loops.
On a report by Import AI
AIpressr
Breaking News, Properly Distributed
Put your launch where the AI industry breaks its news. Trusted by AI journalists, backed by AI founders.
Submit a release →
Skill required for LLMs reportedly underestimated
Timothy B. Lee argues that mastering LLMs involves a significant learning curve, contrary to popular belief.
On a report by Simon Willison
Microsoft Research uses AI to map brain responses to language
New method combines LLMs and fMRI to test hypotheses about brain activity.
On a report by Microsoft Research
Google reportedly loses top AI researchers to Anthropic and OpenAI
Key AI talent departs Google for rivals Anthropic and OpenAI, signaling a potential shift in the AI talent landscape.
On a report by TechCrunch AI
Hugging Face launches benchmark for real-world ASR performance
The FFASR Leaderboard aims to quantify the gap between clean-speech benchmarks and far-field ASR accuracy in noisy environments.
On a report by Hugging Face Blog
AI models reportedly struggle with role confusion in text prompts
New research suggests AI models prioritize text style over content, leading to potential security vulnerabilities.
On a report by Simon Willison
AI reportedly outperforms humans in persuasive text-based interactions
New research suggests AI systems are more effective than humans in changing opinions and securing donations.
On a report by Import AI
Nobel laureate John Jumper departs DeepMind for Anthropic
John Jumper, a key figure behind AlphaFold, leaves DeepMind after nearly a decade to join rival Anthropic.
On a report by TechCrunch AI
AI research agents reportedly leak sensitive enterprise data
Hugging Face Blog study finds training for task performance increases privacy risks in deep-research agents.
On a report by Hugging Face Blog
LoRA dominance questioned in parameter-efficient fine-tuning debate
Hugging Face explores whether LoRA remains the best choice for fine-tuning AI models efficiently.
On a report by Hugging Face Blog
Google's AMIE AI reportedly matches doctors in disease management
Google's medical AI system AMIE shows potential in long-term health condition management, according to new research.
On a report by Google AI Blog
AI founder argues for deep-tech investment over surface products
Alexander Kardos-Nyheim highlights the risks of building AI applications without foundational innovation.
On a report by Crunchbase News AI
DeepMind outlines safeguards for advanced AI systems
DeepMind proposes a layered security framework to manage risks from increasingly autonomous AI agents.
On a report by DeepMind Blog
Anthropic reportedly tests AI model for cybersecurity defense
Anthropic shares White House report on Fable jailbreak with cybersecurity expert Katie Moussouris.
On a report by Simon Willison