1. China's Z.ai GLM-5.2 matches Mythos on cybersecurity — Zhipu AI's open-weight model matches Anthropic's Mythos on bug-finding and cybersecurity, narrowing the US-China AI gap. The Verge · Jun 28
2. Google caps Meta's Gemini usage amid AI infrastructure crunch — Google can't provide the cloud capacity Meta and other clients need, revealing infrastructure bottlenecks despite tens of billions in data center investment. The Verge / Financial Times · Jun 28
3. Ford rehires 'gray beard' engineers after AI falls short — Ford brought back retired engineers when automated quality systems made mistakes, proving domain expertise remains essential alongside AI. TechCrunch · Jun 28
4. Suno launches Spark incubator with controversial gag order — Grants for independent artists paired with a "Good Vibes Only" non-disparagement clause, broad IP license, and class-action waiver. The Verge · Jun 28
5. Prosecutors used ChatGPT logs as evidence in Palisades fire trial — One of the first times AI chatbot conversations have been introduced as evidence in a criminal proceeding. The Verge · Jun 28
6. Wall Street: Micron could be 'the next Nvidia' — Analysts bet that AI's insatiable demand for memory chips will drive Nvidia-style growth for the US memory maker. TechCrunch · Jun 28

---

🧪 Try This Next

• 🔧 Download GLM-5.2 from Hugging Face — run it locally to see what an open-weight model that matches Mythos on cybersecurity feels like
• 🎥 Study Ford's "gray beard" approach — pairing automation with experienced human judgment is a design pattern, not a step backward
• 📝 Read Suno Spark's terms — they're public and a masterclass in what not to sign as a creator