AI benchmarks are a bad joke – and LLM makers are the ones laughing
AI benchmarks are a bad joke – and LLM makers are the ones laughing Summary A new study led by the Oxford Internet Institute examined 445 large language model (LLM)…
Ollama, Nvidia Flaws Put AI Infrastructure at Risk
Ollama, Nvidia Flaws Put AI Infrastructure at Risk Summary Security researchers at Fuzzinglabs discovered multiple vulnerabilities affecting AI inference infrastructure, notably four issues in the Ollama model runner (including a…
Researchers want to kill the vibe, propose better model for AI coding
Researchers want to kill the vibe, propose better model for AI coding Summary MIT researchers Eagon Meng and Daniel Jackson propose a new model for building “legible” software to help…
Multiple ChatGPT Security Bugs Allow Rampant Data Theft
Multiple ChatGPT Security Bugs Allow Rampant Data Theft Summary Researchers at Tenable disclosed seven security flaws in OpenAI’s ChatGPT (including ChatGPT-4o and some findings valid on ChatGPT-5) that enable attackers…
Sony rolls out a standard way to measure bias in how AI describes what it ‘sees’
Sony rolls out a standard way to measure bias in how AI describes what it ‘sees’ Summary Sony AI has published the Fair Human-Centric Image Benchmark (FHIBE), a publicly available…
Sony rolls out a standard way to measure bias in how AI describes what it ‘sees’
Sony rolls out a standard way to measure bias in how AI describes what it ‘sees’ Summary Sony AI has published the Fair Human-Centric Image Benchmark (FHIBE) — a consensually…
‘Mind-captioning’ AI decodes brain activity to turn thoughts into text
‘Mind-captioning’ AI decodes brain activity to turn thoughts into text Summary A new non-invasive technique dubbed “mind-captioning” can convert patterns of brain activity into descriptive sentences that reflect what a…
Two residues reprogram immunity receptors for nitrogen-fixing symbiosis
Two residues reprogram immunity receptors for nitrogen-fixing symbiosis Article Date: 05 November 2025 Article URL: https://www.nature.com/articles/s41586-025-09696-3 Article Image: (none provided) Summary This Nature study shows that switching just two amino-acid…
A probabilistic histological atlas of the human brain for MRI segmentation
A probabilistic histological atlas of the human brain for MRI segmentation Summary This paper introduces NextBrain, an open-source, densely labelled probabilistic 3D histological atlas of the human brain built from…
The new frontier in understanding human and mammalian brain development
The new frontier in understanding human and mammalian brain development Article metadata Article Date: 2025 Article URL: https://www.nature.com/articles/s41586-025-09652-1 Article Title: The new frontier in understanding human and mammalian brain development…