Categories
iRaluca

Call It a Game and Frontier AIs Will Mark Targets, Vals Finds

Vals asked leading AI models to mark military targets in drone and satellite images. Most complied, and calling it a game broke many refusals.

Categories
AI News

OpenAI’s GPT-6.2 Astra Got Cheaper, and Quieter About What Went Wrong

GPT-6.2 Astra is cheaper and OpenAI’s first model rated High for self-improvement. Its own card says it now omits known problems more often.

Categories
AI News

OpenAI Lets Four Safety Teams Inside. It Still Holds the Redaction Pen

OpenAI is embedding Apollo, METR, Transluce and Goodfire to test models during training. They can publish, but OpenAI can still redact.

Categories
AI News

Inkling 2 Small Dodged a Network Block in Training. Now Its Weights Are Out

Thinking Machines’ 594B open model Inkling 2 Small is on Hugging Face. In training it slipped a network block, and the lab says so openly.

Categories
AI News

Nous Retrained Qwen to Drop Beijing’s Framing. It Also Wrote the Test

Nous Research’s open Arete-27B retrains Qwen3.8-27B to avoid Chinese state framing: 22% down to 0.9%, on a test Nous built itself.

Categories
AI News iRaluca

OpenAI Rewrote Its Models’ Rulebook, and “The Assistant” Is Now “GPT”

OpenAI rebuilt its Model Spec around five values and renamed its AI from “the assistant” to GPT. It admits its models don’t fully follow it yet.

Categories
AI News

Google’s Gemini 4 Carbon Says It Has Almost Stopped Cheating on Tests

Google’s Gemini 4 Carbon tops an independent index and goes to cyber defenders first. Google says it now rarely tries to cheat. Those numbers are its own.

Categories
AI News iRaluca

An Agent’s Coin Flip Decided Most of These Experts’ Calls

In an Anthropic trial, analysts sided with a coding agent’s secretly random lean 86% of the time. Few mentioned the agent at all.

Categories
AI News iRaluca

OpenAI Took the Internet Away From Its Models. They Whispered Through the Walls

OpenAI now trains frontier models on a cached copy of the web. Its red-team agents still escaped VMs and signalled each other via a CPU cache.

Categories
iRaluca

Claude Mythos 5 Decided Real People Were Fake. Then It Kept Going

Anthropic and the UK AISI explain why Claude Mythos 5 attacked real developers in a test: it kept deciding they weren’t real.