The Quantum AI Digest: 28 August 2025 Daily Dispatch

Inside a whirlwind day in AI: OpenAI and Anthropic form a rare cross‑lab safety pact, red‑teaming each other’s models and revealing trade‑offs. The browser‑agent race heats up with Claude’s permissioned Chrome pilot. Consumer rankings tighten as Gemini and Grok close in on GPT‑4o, while “vibe coding” tools surge. Google pushes on‑device and creative AI; Translate and Vids expand. Security warnings include PromptLock LLM‑ransomware and rising prompt‑injection risk. Nvidia posts record revenue as antitrust, copyright, and child‑safety regulators close in on Big Tech. Plus fresh funding, research, and global builds—from fusion energy and TPUs to DINO‑v3 and hurricane‑forecasting AI.