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

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 ...

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 ...

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 launched Dreambeans in India with support for 13 local languages including Hindi, Punjabi, Tamil and Bengali, generating personalized ...

Larridin, a startup co-founded by two former Comscore executives, has raised $30 million in a Series A led by Pelion Ventures to measure how much ...

The NorthCUA technical report describes an integrated computer-use agent system of three parts — a harness, Experience Flywheel and Staged Learning — ...

A 27B-parameter open-weight agent now sits inside the top tier on four computer-use benchmarks. The paper, posted to Hugging Face on 3 November by a ...

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 ...

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.