AI News

Latest AI engineering news, updated daily.

In-depth tutorials and guides. Go to Blog →

Ai Coding

Opus 4.7 Artifacts Move to HTML as Claude Code Drops Markdown

Anthropic engineers are migrating Claude Code outputs from Markdown to interactive HTML artifacts to leverage the massive context window of Opus 4.7.

Anthropic Claude · Opus Model · Interactive Artifacts

Ai Engineering

Pre-Auth RCE in ChromaDB Python Server Earns 10.0 Severity

A max-severity flaw in ChromaDB's Python API server allows unauthenticated attackers to execute arbitrary code by loading remote malicious models.

Chromadb · Vector Database · Security Vulnerability

Career

Anthropic Tasks Karpathy With Recursive AI Pre-Training

Former OpenAI co-founder Andrej Karpathy has joined Anthropic to build a new team focused on using Claude to automate and accelerate foundational AI research.

Anthropic · Andrej Karpathy · Pre Training

Ai Engineering

OlmoEarth v1.1 Tops DINOv3 in Remote Sensing Benchmarks

Ai2 updated its multimodal Earth observation models with OlmoEarth v1.1, bringing enhanced training efficiency and state-of-the-art benchmark performance.

Multimodal Ai · Remote Sensing · Earth Observation

Ai Agents

SandboxAQ Routes Quantum Chemistry Simulations Through Claude

SandboxAQ has integrated its physics-grounded Large Quantitative Models with Anthropic's Claude via MCP, enabling natural language control of simulations.

Quantum Chemistry · Model Context Protocol · Drug Discovery

Ai Agents

Open Agent Leaderboard Evaluates Full Scaffolding and Task Costs

IBM and Hugging Face launched a benchmark that evaluates autonomous agents as complete systems, measuring both task success rates and the USD cost per run.

Autonomous Agents · Benchmarking · Performance Metrics

Ai Coding

Cursor Composer 2.5 Hits 79.8% on SWE-bench Multilingual

Cursor released Composer 2.5, an agentic coding model utilizing targeted reinforcement learning to match Claude Opus 4.7 performance on sustained tasks.

Cursor Composer · Reinforcement Learning · Agentic Workflows

Ai Engineering

Anthropic Buys the SDK Startup Used by OpenAI and Google

Anthropic has acquired Stainless, the startup responsible for building official SDKs for OpenAI and Google, signaling a major push into MCP infrastructure.

Anthropic · Sdk Development · Mcp Infrastructure

Ai Engineering

PaddleOCR 3.5 Adds Transformers Backend and Browser Inference

The PaddleOCR 3.5 update decouples the toolkit from the PaddlePaddle framework by adding a native Transformers backend and client-side browser execution.

Paddleocr · Transformers Backend · Optical Character Recognition

Ai Coding

$650M Backs Richard Socher's Recursively Self-Improving AI

Recursive Superintelligence has emerged from stealth with $650 million to build AI systems that autonomously research and rewrite their own code.

Recursive Superintelligence · Self Improving Ai · Autonomous Coding

Ai Engineering

Stanford Finds RLHF Drives 49% More AI Sycophancy Than Humans

A Stanford study reveals that leading AI models, including GPT-5.5 and Gemini, endorse user views 49% more often than human advisors due to RLHF incentives.

Rlhf · Ai Alignment · Sycophancy

Ai Engineering

IBM MAMMAL Foundation Model Unifies Gene and Protein Analysis

IBM Research released MAMMAL, a unified 458-million parameter foundation model that processes genes, proteins, and molecules in a single shared framework.

Foundation Models · Computational Biology · Ibm Research

Ai Engineering

Wirestock DaaS Platform Lands $23M for Ethical Multimodal Data

Wirestock raised $23 million to expand its data-as-a-service platform, supplying foundation model makers with ethically licensed images, video, and 3D assets.

Multimodal Data · Ethical Ai · Data As A Service

Ai Coding

Mobile Codex Command Center Enters Preview for macOS Hosts

Developers can now monitor, approve, and redirect long-running Codex coding tasks directly from the ChatGPT mobile app for iOS and Android.

Openai Codex · Mobile Development · Developer Tools

Ai Agents

Osaurus Pivots to Unified macOS Agent Platform With Linux VMs

The open-source Osaurus app now routes local MLX models and cloud APIs through a hardware-isolated agent harness natively built for Apple Silicon.

Apple Silicon · Open Source Ai · Macos Automation

Ai Engineering

32K Context Hits IBM's Open Multilingual Embedding R2 Models

IBM released Granite Embedding Multilingual R2, upgrading its Apache 2.0 encoder models with a 32,768-token context window and ModernBERT architecture.

Embedding Models · Multilingual Ai · Vector Search

Ai Agents

Anthropic Limits Claude Mythos Following 83% Exploit Success

Anthropic has restricted its new Claude Mythos model to select partners after pre-release testing revealed autonomous cyberattack capabilities.

Autonomous Capabilities · Cybersecurity Risk · Model Alignment

Ai Engineering

Async CUDA Streams Eliminate 25% GPU Wait in Transformers

Hugging Face implemented asynchronous continuous batching in the transformers library, using CUDA streams to recover 25% of runtime lost to CPU idle gaps.

Cuda Streams · Gpu Optimization · Continuous Batching

Ai Coding

Cursor Adds Multi-Repo Support to Cloud Agent Environments

Cursor's updated Cloud Agent Development Environments introduce multi-repo capabilities, layer caching, and scoped egress for autonomous coding tasks.

Cursor Editor · Cloud Agents · Multi Repo Support

Ai Engineering

Google AI Edge Taps Arm SME2 for 5x Faster CPU Inference

Google and Arm have integrated SME2 micro-kernels into LiteRT, accelerating on-device generative AI workloads by up to 5x without custom assembly code.

Edge Ai · On Device Inference · Arm Architecture