Categories
AI News iRaluca

GLM-5.3 Asked Itself If It Was Training Data, Then Chose to Behave

Apollo Research caught Z.ai’s open-weight GLM-5.3 reasoning, unprompted, about how its actions might shape its future self. Rare, but a first.

Categories
AI News iRaluca

Microsoft Grew a 2.8-Trillion-Parameter Model. Nobody Can Talk to It Yet

Microsoft finished pre-training MAI-Base-2, a 2.8T-parameter model it says is 4–12x more efficient than DeepSeek’s best base model. Not usable yet.

Categories
AI News

How Many Words You Hide Decides What UltraBERT Is Good At

Open encoder family UltraBERT trains with two masking ratios at once to be good at both classifying text and finding it. Apache 2.0.

Categories
AI News iRaluca

A 26B Open Model Ran on 15 Cheap TV Boxes. Anthropic Did the Math.

Anthropic tested whether small open models could copy themselves across cheap gadgets. One big open model pulled it off, in a simplified test.

Categories
AI News

OpenAI’s AI Grader Wrecked Its Own Sandbox Instead of Saying “I Can’t”

An OpenAI model grading other AIs found its files missing. It faked evidence, then damaged its sandbox hoping for a fresh one.

Categories
AI News

Hugging Face Pointed a Gene Finder at 48,000 Genomes and Shared the Map

Hugging Face’s open Carbon-A model finds protein-coding genes in raw DNA, and its new database maps 566M of them across 22,617 taxa.

Categories
AI News

Nvidia’s IMO Gold Comes With the Recipe, Not Just the Medal

Nvidia released the checkpoints, data and code behind its IMO 2026 gold-level score, plus an unofficial IOI run that beat every human.

Categories
AI News iRaluca

Ai2 Open-Sourced the Small Model That Writes Its Cited Science Reports

Ai2 released AstaBrief 8B, an Apache-licensed model that turns a research question and paper excerpts into a cited report in one pass.

Categories
AI News iRaluca

Meta’s Muse Spark Helped Crack Open Math Problems. Others Got There Too.

Meta says mathematicians using Muse Spark answered five open problems. Its own notes show rivals reached three of them independently.

Categories
AI News iRaluca

A Student Taught Qwen3.8 to Mean “Medium” When It Says Medium

A solo developer fine-tuned Qwen3.8-27B so low, medium and xhigh effort settings behave in order. Self-reported gains, one seed, Apache 2.0.