AI News
Latest AI engineering news, updated daily.
Ai Engineering
Shrinking Model VRAM by 22% with Cloudflare Unweight
Cloudflare's new Unweight system offers lossless, bit-exact LLM compression, saving 3GB of VRAM on 8B models without impacting output quality.
Cloudflare · Unweight · Llm Inference
Ai Agents
Empowering AI Agents With Cloudflare Email Service Beta
Cloudflare launches its Email Service public beta, enabling AI agents to natively send, receive, and process emails with integrated security and MCP support.
Cloudflare Workers · Email Sending · Ai Agents
Ai Engineering
Boosting Kimi K2.5 Speed 3x via Cloudflare Infire Optimization
Cloudflare enhances Workers AI with the Infire engine, enabling extra-large models like Kimi K2.5 to run faster and more cost-effectively using Rust-based optimizations.
Cloudflare Workers · Kimi K2 5 · Llm Inference
Ai Agents
Scaling AI Agent Workflows with Git-Compatible Artifacts
Cloudflare launches Artifacts, a Git-powered versioned filesystem designed to handle millions of repositories for autonomous AI agent workloads.
Cloudflare · Artifacts · Git Native
Ai Coding
Claude Opus 4.7: Better Coding, 3x Vision, Cyber Controls
Anthropic releases Claude Opus 4.7 with major software engineering gains, 3x higher image resolution, automated cybersecurity safeguards, and a new xhigh effort level.
Claude Opus · Anthropic · Software Engineering
Ai Coding
Codex Can Now Use Your Mac While You Work
OpenAI's Codex desktop app adds background computer use on macOS, an in-app browser, image generation via gpt-image-1.5, and 90+ new plugins.
Openai Codex · Computer Use · Ai Coding
Ai Engineering
MoGen Synthetic Data Slashes Brain Mapping Error Rates
Google Research debuts MoGen, a generative model creating synthetic neurons to save 157 person-years of manual proofreading in mouse brain reconstruction.
Mogen · Connectomics · Google Research
Ai Engineering
Nano Banana 2: Putting Your Own Photos Inside Gemini Images
Google's Nano Banana 2 powers personalized image generation in the Gemini app, letting users include themselves and loved ones using their Google Photos library.
Gemini App · Nano Banana 2 · Google Photos
Ai Engineering
Google’s Simula: Architecting Datasets via Mechanism Design
Google Research introduces Simula, a reasoning-first framework that treats synthetic data generation as programmable mechanism design for better model training.
Synthetic Data · Google Research · Simula
Ai Agents
Untrained Tasks Now Possible via π0.7 Robotic Brain
Physical Intelligence unveils π0.7, a foundation model enabling robots to solve novel, complex problems through compositional generalization.
Physical Intelligence · Robotics · Foundation Models
Ai Engineering
GPT-Rosalind: OpenAI's New Model Outperforms Human Experts
OpenAI's GPT-Rosalind is a specialized life sciences reasoning model targeting drug discovery, genomics, and protein engineering, with a free Codex plugin for tool integration.
Gpt Rosalind · Life Sciences · Drug Discovery
Ai Agents
Agentic Creativity: Adobe Firefly AI Assistant Automates Apps
Adobe's Firefly AI Assistant acts as a cross-application agent to automate complex creative workflows across Photoshop, Premiere Pro, and Illustrator.
Adobe Firefly · Creative Cloud · Ai Assistant
Ai Coding
Beyond Text: Cursor Canvases Render Agent Data Visually
Cursor introduces Canvases, enabling AI agents to generate interactive React-based UI interfaces and dashboards for complex data visualization and debugging.
Cursor Editor · Canvases · Ai Agents
Ai Agents
AI Agents Get Post-Quantum Networking in Cloudflare Mesh
Cloudflare Mesh introduces a secure fabric for AI agents, users, and nodes, replacing legacy VPNs with identity-based, post-quantum encrypted connectivity.
Cloudflare Mesh · Ai Security · Post Quantum Encryption
Ai Agents
Cloudflare released Managed OAuth to Secure AI Agents
Cloudflare Access introduces Managed OAuth to help AI agents securely authenticate with internal apps using RFC 9728 and dynamic client registration.
Cloudflare Access · Managed Oauth · Ai Security
Ai Agents
Why AI Agents Still Fail at Complex Tasks
A new IBM Research analysis explores the VAKRA benchmark, revealing how top AI models struggle with multi-hop reasoning and live API chaining in enterprise tools.
Ibm Research · Vakra Benchmark · Api Chaining
Ai Engineering
Claude Mythos Can Hack Autonomously, and the UK Is Worried
The UK government urges businesses to harden defenses after Anthropic's Claude Mythos Preview demonstrates unprecedented autonomous hacking and zero-day discovery.
Anthropic · Claude Mythos · Cybersecurity
Ai Engineering
A Critical Nginx UI Flaw Is Being Actively Exploited
The 'MCPwn' flaw (CVE-2026-33032) allows unauthenticated attackers to hijack Nginx UI instances via a neglected AI protocol endpoint.
Nginx Ui · Cybersecurity · Vulnerability Alert
Ai Coding
Windsurf 2.0 Is Now a Multi-Agent Command Center
Codeium’s Windsurf 2.0 transforms the AI IDE into a multi-agent orchestrator, allowing developers to manage Devin and Cascade agents from a central hub.
Windsurf · Devin · Agentic Workflows
Ai Engineering
Gemini 1.5 Flash Now Does Real-Time Voice
The new Multimodal Live API enables developers to build low-latency, expressive speech-to-speech applications with advanced emotional inflection.
Google Gemini · Speech To Speech · Multimodal Ai