Sarah and Alex break down the latest AI headlines: MLB cracks down on AI pitch calls, China's Kimi K3 goes open-source, SpaceX eyes Pentagon cloud work, a humanoid robot loses its head in the ring, hackers weaponize AI tool searches, AI-run companies buy real land, and the no-code billionaire argument heats up.

Apple tests PrismML compression that runs 27 billion parameters on an iPhone. Suno leaks reveal training sources and customer data. OpenAI's ad forecast looks off, AI-built design still feels generic, and a self-hosted coding assistant emerges as a Copilot alternative.

OpenAI is reportedly building a screenless speaker and shipping a mechanical keyboard for Codex agents. Robots performed the world's first remote surgery, Google added image generation to Search, and researchers found a single string that freezes AI hackers. Plus a neuroscientist's case against AI consciousness, and open-source tools for model routing and on-device voice cloning.

ChatGPT is back on WhatsApp, putting AI in front of two billion users. Sixteen Nobel laureates warn of a jobs tsunami unfolding in just ten years. Plus: Claude's values shift by language, Proton launches a fully private AI chat, and Waze adds motorcycle mode with Gemini.

GPT-5.6 multi-agent architecture. Claude usage dashboard. Companion robots mass production. Smart toothbrush with AI. Google AI certifications. Ancient Greeks meet AI. New creative tools.

Meta reverses its likeness tool after backlash. Apple sues OpenAI over trade secrets. Brown professor busts 96%% of AI cheaters. Ghost Font fools AI. Google Edge Gallery. Microsoft free AI course. Roadmap to superintelligence.

Lifelike companion robots enter mass production with on-device emotion processing. Meta's first custom AI chip enters production while Ollama raises 65 million dollars. Plus: GPT-5.6 wins its first ARC-AGI game, Google launches a free AI certificate, AI invents new molecules, and a pocket E Ink reader for capturing stray ideas.

OpenAI ships GPT-5.6 with a coordinated multi-agent architecture using Sol, Terra, and Luna, alongside the new ChatGPT Work productivity app. Anthropic gives Claude a usage dashboard, Meta rolls out Muse Spark 1.1 with agent capabilities, and an open-source TTS model runs entirely on CPU.

OpenAI turns ChatGPT into a real-time conversational voice with GPT-Live, Grok ships 4.5 plus 21 multilingual voices, and .lumen glasses aim to replace guide dogs with camera-based navigation. We also cover Cloudflare's drag-and-drop hosting, the argument that enterprise AI is still less than one percent built, and why the ancient Greeks may have seen AI coming.

DeepSeek is reportedly designing its own AI inference chip. Anthropic is bringing Claude Cowork to mobile and web, while Meta launches its world-aware Muse Image generator inside Meta AI. Humanoid robots are now rentable by the day, the synthetic actor Tilly Norwood lands a lead feature-film role, AI financial advice varies by race and gender in a new study, and Feno's twenty-second smart toothbrush scans your mouth while it cleans.