Meta Confirms Serious Security Incident Caused by Rogue AI Agent

The AI coding assistant landscape is heating up with Cursor's launch of Composer 2, its new in-house coding model. According to benchmark results released by the company, Composer 2 has outperformed Anthropic's highly-regarded Claude Opus 4.6 on key coding metrics, though it still trails behind OpenAI's top-tier GPT-5.4 model. The new model is integrated directly into Cursor's agentic coding environment, which is designed to act more like an autonomous programming partner than a simple code-completion tool. This release signals Cursor's ambition to compete at the highest level of AI-assisted software development, offering developers improved code generation, understanding, and problem-solving capabilities. By building its own specialized model, Cursor aims to provide a more tightly integrated and performant experience within its ecosystem, challenging the dominance of general-purpose large language models in the coding domain.