AI news today: safety, devices, superapps, and model-race drama

AI's biggest stories today aren't model demos. They're safety tension, hardware moves, superapp plans, and model-theft allegations. Here's the 4-item stack Reuters is pointing to, July 24, 2026^[1][2][3][4].

Reuters says a US-China feud is threatening AI safety efforts, a sign that governance is getting harder just as the tech gets more powerful, July 24, 2026^[5].

OpenAI is reported to have 200+ people on AI devices, including a smart speaker and possibly smart glasses and a smart lamp. That's software moving into hardware, Reuters, July 24, 2026^[6].

OpenAI is reportedly planning its biggest ChatGPT overhaul yet, turning it into a superapp with coding tools and AI agents to boost revenue before a possible stock listing, Reuters, July 24, 2026^[7].

Allegation alert: Reuters says China's Moonshot AI allegedly stole from Anthropic's Fable to help build its latest K3 model, and also reportedly got advanced Nvidia chips. That keeps the IP and compute race hot, July 24, 2026^[8][9].