MCP Protocol: Design flaw puts 200,000 servers at risk
Security researchers found a critical vulnerability in Anthropic's Model Context Protocol. 200,000 servers affected, 150 million downloads compromised. Anthropic calls the behavior 'expected'.
Claude Code Gets a Desktop Overhaul and 'Routines' — Your AI Agent Now Works While You Sleep
Anthropic completely rebuilt Claude Code's desktop app with parallel sessions, an integrated terminal, and a brand-new Routines feature that runs your AI tasks automatically in the background.
Is Claude Getting 'Nerfed'? Power Users Push Back Against Anthropic's Effort Level Change
Developers and heavy users are reporting noticeable quality drops in Claude. The reason: Anthropic quietly lowered the default effort level to 'medium' — and the community is not happy about it.
Claude Sonnet 4 and Opus 4 Retire on June 15 — What You Need to Do Now
Anthropic announced the deprecation of the original Claude 4 models. Starting June 15, 2026, API calls to Sonnet 4 and Opus 4 will fail. The migration is straightforward.
OpenAI Launches GPT-5.4-Cyber: An AI Model Reserved for Security Professionals
OpenAI released a specialized cybersecurity model available only to vetted security researchers. GPT-5.4-Cyber can reverse-engineer binaries and find vulnerabilities.
Gemini 3.1 Flash TTS: Google's New Speech Model Speaks 70 Languages — and Takes Direction
Google released a text-to-speech model you can control with prompts. From 'enthusiastic' to 'Brixton accent' — it's all there.
Gemini Now Knows Your Face: Google Brings Personalized Image Generation with Nano Banana
Google connects Nano Banana to your photo library. Say 'make a picture of me as an astronaut' — and Gemini knows what you look like.
OpenAI Codex Gets Desktop Superpowers: Background Agents, Browser, and 111 Plugins
OpenAI just shipped a massive Codex update — the desktop app can now control other programs, generate images, and has a built-in browser. A direct shot at Claude Code.