Fine-tuning on benign-looking data can quietly erode safety guardrails in ways standard content filters cannot detect.

Want only the AI that hits your stack? Build an agent that tracks your companies and topics, and skips everything else. A 27B-parameter open-weight agent now sits inside the top tier on four ...

Embedding research across text, code, image and video, audio, graph, table and time series is quietly converging on a shared recipe even while ...

"Fine-tuning Large Language Models (LLMs) on instruction data can quietly erode their safety, even when no training example looks harmful." That is ...

We scan all of AI. You track a slice of it. Build an agent that watches only the companies, topics and people you choose. It reads everything our scanners pull and sends you an edition made of just ...

Tsinghua University and UC Berkeley introduce a modular uni-robot-agent harness built around open models. Paired with Qwen3.8-Flash-Next, it scores ...

Build an agent that watches only the companies, topics and people you choose. It reads everything our scanners pull and sends you an edition made of just your matches.

A solo researcher trained two competitive mixture-of-experts models from scratch using roughly a 12th of the median training data of their ...

We scan all of AI. You track a slice of it. Build an agent that watches only the companies, topics and people you choose. It reads everything our scanners pull and sends you an edition made of just ...

Google has launched Dreambeans in India, bringing its proactive AI story app to 13 local languages including Hindi, Punjabi, Tamil, Telugu, and ...

Want only the AI that hits your stack? Build an agent that tracks your companies and topics, and skips everything else. Larridin, a startup co-founded by two former Comscore executives, has raised $30 ...

Nvidia's two newest humanoid reference platforms both ship with robot hands from China-linked startups. CNBC's China Connection newsletter reports ...