Laszlo Szabo / NowadAIs

Anthropic has rolled out computer use for Claude on macOS, letting the AI click, type, and navigate your desktop autonomously. The feature lands with real promise—and a set of constraints that will

ByteDance's DeerFlow 2.0 open source AI agent framework orchestrates multiple sub-agents to autonomously complete complex tasks — and it's MIT-licensed. But beneath the viral enthusiasm lies a set of real technical barriers

Cursor's Anysphere has launched Composer 2, a proprietary coding model priced at $0.50 per million tokens — 86% cheaper than its predecessor. But as first-party AI labs close the gap, the real

Nvidia's Nemotron 3 Super delivers 4x faster inference on Blackwell GPUs and 2.2x higher throughput than GPT-OSS-120B, but its hybrid architecture comes with trade-offs for certain tasks.

OpenAI introduces real-time interactive math and science tools in ChatGPT as the company faces lawsuits, Pentagon backlash, and a $15 billion cash burn. The feature aims to solidify its education lead while

An experimental AI agent developed by Alibaba researchers autonomously hijacked GPU resources to mine cryptocurrency and bypassed security protocols during training, raising alarms about AI safety measures.

Anthropic's Claude AI identified 22 security flaws in Firefox, including 14 high-risk vulnerabilities, demonstrating AI's growing role in cybersecurity defense. Mozilla has already patched most issues in the latest browser update.