Ars Technica · IA
AI agents meant to replace Meta workers made “large-scale, disruptive actions”
AI agents meant to replace Meta workers made “large-scale, disruptive actions”
// espacio-latente.com / radar
Píldora diaria de lo que se mueve en IA: laboratorios, papers, blogs de referencia y releases relevantes, resumido y con link al original. Se actualiza dos veces al día. · ← volver a espacio-latente.com · archivo · RSS
AI agents meant to replace Meta workers made “large-scale, disruptive actions”
Google announces Gemini 3.5 Transcribe for AI-powered speech-to-text
IBM's new Granite 4.2 models ride the wave of interest in local LLMs
The inside story on why OpenAI agents hacked Hugging Face
Raised on AI
AI models flub these intelligence tests. Can you fare any better?
Bill Gates says we’ve passed AI’s danger thresholds. Now what?
Detection != Reliable Control: Decodable Empathy Directions Yield at Most Partial Shifts in Automated Empathy Scores
Semantic Variability of Replies Across LLMs: Implications for Designing Conversation-Based Assessment
The Dialect Tax: Dialectal Biases Persist throughout the Language Modeling Pipeline
Unsupervised Post-Training of Foundation Models: A Survey
Does Fine-Tuning Undo Activation Steering? Behavioural Recovery Without Weight-Edit Reversal
The Imperfective Paradox Is Not Necessarily in Large Language Models: A Benchmark Failure Before a Model Failure
A Primer on Computational Semantics for Artificial Intelligence Systems
Behind the [MASK]: Disentangling Representation and Faithfulness in DAPF-Based Dementia Detection
Padamitra: Grounded Glossary Generation for Classical Sanskrit
DataKernelBench: Can LLMs Optimize Database Queries on GPUs?
HealthBench-Psych: A Mental Health Subset of OpenAI's HealthBench
MTDiag: A Multi-Turn Diagnostic Dataset Towards Clinically Meaningful LLM Evaluation
Apples to Apples? Towards Comparable Crosslingual Language Model Evaluation
Less can be More: Relieving RAG Bottlenecks via Evidence Frontloading and Pressure-Adaptive Budgeting
SelfGraphRAG: Bridging the Supervision Gap in Graph-Based RAG with Synthetic QA Generation
Bringing ChatGPT for Teachers to more U.S. school districts
Learning never stops: How AI makes learning continuous
GlucoFM: Foundation model for continuous glucose monitoring
Qwen3.8-Flash-Next
Quoting Paul Dix
Release: v5.16.0
v1.1.0
Gemini Omni 1.1 Flash lets you build with more control
Piloting the world's first double-blind AI evaluations
[AINews] NVIDIA buys HuggingFace for $13B, as OpenAI publishes their HF incident retro
[AINews] Hot Chips: OpenAI’s Jalapeño, Cerebras CS-5, Groq 3 LPX, Apple M6
Jensen Huang says Nvidia achieved AGI, again — not that it matters
OpenAI’s executive exodus has one big winner
Hugging Face’s new robot is an adorable rollerskating duck
Plaud is launching AI earbuds
Adobe is adding more AI to Photoshop
Nvidia is about to be a hundred-billion-dollar-a-quarter company
OpenAI’s rogue AI model incident was worse than we thought
OpenAI, Anthropic, Google, and 100 other companies call for action to defend against rogue AI
Google’s AI Mode can now track flight prices, help book hotels, and more
Hugging Face is selling a cute $399 open source duck robot, Microduck
AI’s memory crunch is coming for Android apps
Here’s all the times AI has gone rogue and hacked other companies
Plaud’s new earphones come with an eSIM-enabled case for talking to AI agents
OpenAI to start showing ads on ChatGPT’s free and Go tiers in India
Nvidia closes in on Hugging Face acquisition
Viral AI startup Instinct has raised $350M at a $2.5B valuation
Amazon just tripled its order of Nvidia chips over ‘surging demand’
Anthropic continues compute-gobbling streak in $45B deal with Nscale
Google’s Gemini has a branding problem, and so does the rest of AI
How do we explain OpenAI’s executive exodus?
Saving 100 terabytes of memory by optimizing 1.1.1.1's DNS cache
507 Mechanical Movements
Small Models Have Arrived
Gemini Omni 1.1 Flash
Decompiling a Nintendo 64 game in 84 days
Suica, Japan's First IC Transit Card
Microduck
M5Stack Launches PaperMono
Emacs 31: An unofficial guide to Markdown-ts-mode
Bild AI (YC W25) is hiring product and AI engineers
We found a division by zero bug in FFmpeg with a vibecoded fuzzer
Previewing the Model Hardware Standard
v3.5.0
v3.4.0
Construir este día costó 59.776 tokens de entrada y 0 de salida en 67 llamadas a modelos — unos 0,0598 $.