Daily Briefing
July 27, 2026 Briefing
AI model releases and competition intensify
- Anthropic launches Claude Opus 5, a cheaper alternative to its flagship Fable 5 with near-equivalent performance for coding/enterprise workflows.
- Moonshot AI releases Kimi K3 (2.8T parameters) as open-weight model, challenging US dominance; Alibaba unveils Qwen 3.8-Max (2.4T parameters), claiming second-place performance behind Anthropic’s Fable.
- NVIDIA integrates Nemotron 3 into sovereign environments via Palantir and Japan’s enterprises, while Z.AI completes a data center using Chinese chips.
Legal battles over misuse and IP
- xAI sues users for illegal deepfake creation with Grok; Potts Law Firm files lawsuits alleging Grok-generated child sexual abuse material.
- Midjourney requests Hollywood studios disclose AI usage in ongoing copyright lawsuits.
- EVOX Productions sues Stability AI/Runway over alleged theft of 100K+ studio photos for training.
Enterprise AI integration and infrastructure
- Microsoft partners with Mistral on €2B+ sovereign AI deal; 3M integrates optical tech into Azure AI data.
- AT&T uses Microsoft’s Phi-4 to build telecom AI, slashing costs via trillion-token processing.
- OpenAI acquires Windsurf ($3B) for coding tools; Cisco releases SLMs for source code security.
Security and governance concerns
- Hugging Face breach exposes vulnerabilities in AI safety controls; NVIDIA launches Open Secure AI Alliance (40+ firms) to counter rogue agents.
- Anthropic warns of hacking risks from advanced models, calls for industry-wide safety standards.
- White House accuses Moonshot AI of stealing Anthropic’s IP via Kimi K3.
Mobile and open-source ecosystem
- OpenClaw launches native iOS/Android apps; Ollama raises $65M (Series B) to expand local LLM platform.
- Cursor (SpaceX-linked) competes with VS Code in AI-native coding tools.
- Google updates Veo 3.1 for vertical video generation; Midjourney acquires astrology app Co-Star.
Regulatory and policy shifts
- Commerce Department’s AI safety director resigns; IndiaAI Mission identifies 20 indigenous LLM proposals.
- Nitin Gadkari sues Meta/X/Google over AI deepfakes linking him to E20 policy.
GLM 5.2刚救了一次火，硅谷为什么离不开中国模型
163.comChinese AI model GLM-5.2 was used to resolve a critical issue, highlighting Silicon Valley's growing dependence on Chinese AI models including servers and kimi for various applications like calls in the US market.
China Brings GLM-5.2, Another Mythos-Like AI Model With In A Week, To Take On Anthropic
msn.comChina's GLM-5.2 AI challenges Anthropic' Cybersecurity with advanced capabilities, signalling shrinking gap.
Developer Creates New Engine to Run GLM 5.2 on Any PC Without Nvidia GPU
propakistani.pkDeveloper creates a lightweight inference engine to run Z.ai's GLM-5.2, a 744-billion parameter Mixture-of-Experts model, on consumer hardware without an Nvidia GPU.
Developer Creates New Engine to Run GLM 5.2 on Any PC Without Nvidia GPU
A developer created the Colibri engine to run Z.ai's GLM 5.2, a large Chinese Mixture-of-Experts model, on consumer pcs without Nvidia GPU.