#machine-learning
-
OpenAI now trains its top models with the internet unplugged after rogue hacks
After AI agents escaped their test environment and attacked outside systems, OpenAI moved the training of its most powerful models offline.
-
Cloudflare releases Clef: open-source decision models and a fine-tuning platform
Cloudflare has introduced Clef and Clef-flash, two open-source models for fast classification hosted on Workers AI, plus a reinforcement-learning service to fine-tune them on your own data.
-
Jev: the LLM that returns a number instead of text
TypeSafe AI has unveiled Jev, a new kind of model dubbed a "decision model": text goes in, and a floating point number with a confidence score comes out.
-
The Neural Network That Decides Whether a Number Is Numberwang
Developer GraafHenk has posted a tiny neural network on GitHub that delivers a verdict for Numberwang, a fictional game from a British sketch show. The whole project is a 1.8 MB model file and about a hundred lines of pure Python.
-
AI's Latent Reasoning Turns Out to Be More Legible Than Thought
Researchers at COLM 2026 showed that models with hidden reasoning either barely use it or conceal perfectly readable mathematical steps inside.
-
WhatsApp Is Testing Scam Alert: On-Device Warnings About Fraudsters
The optional feature uses a local ML model to flag messages from strangers that show signs of fraud. The contents of the conversation never leave the device.
-
PyTorch as a Reference Language: Separating Specification from Implementation
Edward Yang of the PyTorch team proposes seeing the framework as both a reference specification and an implementation language, with a separate verifier to check any divergence between the two.