Large Language Models (LLMs) - The Batch

Four-step process diagram shows misuse of AI models with fake accounts, data harvesting, cleaning, and training.

Machine Learning Research

Some Kimi and DeepSeek Users Were Served Claude Instead, Anthropic Says: Inside Anthropic's accounts of distillation, gray-market transfer stations, straw accounts, and user fraud

Anthropic accused seven Chinese AI developers, including Moonshot AI, Alibaba, and DeepSeek, of misusing its Claude models to mislead their customers and improve their own models.

Flowchart illustrates data process from user control to external services in Meta's Muse AI agent system.

Machine Learning Research

How To Secure Agents for the Masses: Meta's Muse Agent uses your accounts and payments to take action

A malicious web page can fool an AI agent into working against you.

A woman presents LLM safety charts to two colleagues, illustrating CRC Monitor's effectiveness visually.

Machine Learning Research

A Simpler Method to Monitor Models: CRC Monitor can tell if model output is incorrect or unsafe in real-time

LLM safety monitors that act during generation often analyze a series of safety scores to catch incorrect or harmful outputs.

A table compares performance metrics of large language models in legal and general domains.

Machine Learning Research

Custom Models for Law, News, and Finance: Thomson Reuters’ Thomson LLM, based on Qwen 3.5, trained on proprietary data

Thomson Reuters launched a proprietary large language model family called Thomson.

Illustrated process of context management, highlighting context editing before sending to LLM input.

Machine Learning Research

LLMs Take Out the Agent’s Trash: Researchers at Xiaohongshu detail LLM-based memory management technique for agents

AI agents typically compact the contents of their context windows by summarizing or deleting the oldest material.

Plugin configuration screen with multiple plugins enabled, reflecting DeepSeek-V4-Pro’s new features.

Hardware

DeepSeek-V4-Pro Gets Refreshed: DeepSeek’s flagship only slightly outperforms Flash model; Open source harness draws interest

DeepSeek’s flagship model graduated from preview with improved performance plus the harness the model was benchmarked in.

Bar graph compares GPT-5.6 Sol Ultrafast and Sol, highlighting speed differences in processing tasks.

Machine Learning Research

Inside AI’s Need for Speed: OpenAI Partners with Cerebras, Google Releases Gemini 3.7 Flash, Nvidia Debuts Nemotron 3.5 Lightning

Developers who already track models’ cost and accuracy have good new reasons to pay closer attention to a third essential factor: speed.

Visual contrasts error correction between human dialogue, traditional ASR, and interactive ASR paradigms.

Machine Learning Research

Agents Come to Speech Recognition: AgenticASR incorporates user corrections to edit speech recognition on the fly

Most speech-to-text systems transcribe speech in a single pass, which makes them unable to correct errors in their outputs. Researchers built a system that allows for interactive corrections.

Charts show Muse Spark 1.2 scoring lower but more cost-effective, supporting Meta’s low-cost coding model.

Machine Learning Research

Muse Code Wants Your Data: Meta's Muse Spark 1.2 and Muse Code approach the intelligence frontier at a discount

Meta will cut coding bills from dollars to pennies for developers who let the company learn from their work.

Steps detail generating secure prompts using MCMC and rephrasing to minimize Python code vulnerabilities.

Machine Learning Research

Custom Prompts for Safer Code: A Stanford team built a pipeline to improve system prompts to build more secure code

Large language models (LLMs) can write useful code, but they often introduce security vulnerabilities.

The image depicts a chart of token increases in The Stack v3, emphasizing C++ and HTML's significant growth.

Machine Learning Research

Fresh Data for Coding Models: Hugging Face updated The Stack, the largest crawl of public GitHub repositories

The biggest open dataset of source code went years without an update.

A chart shows DeepSeek-V4-Flash leading in a green area with low cost and strong performance, marked in blue.

Machine Learning Research

DeepSeek Pushes the Frontier Again: DeepSeek refreshed its V4 Flash model with an impressive fine-tune

DeepSeek’s updated small model overtook the company’s own flagship.