Here’s what’s worth your time today.
OpenAI has missed internal targets for both user growth and revenue, raising concerns inside the company about whether its aggressive spending on data centers is sustainable. After committing hundreds of billions to secure compute, leadership is now under pressure to rein in costs as growth slows and competition from players like Google and Anthropic intensifies. The tension is clear: scale still depends on owning massive infrastructure, but the economics are starting to bite. With a potential IPO on the horizon, OpenAI is being forced to shift from pure expansion mode to something closer to financial discipline.
Supported by Bland AI
Norm lets you one-shot prompt a fully structured voice agent. Just describe what you want—appointment scheduling, customer support, lead qualification—and watch Norm build the agent, pathways, and logic for you. Safe branching. Agent-on-agent testing. Deploy when you’re ready.
OpenAI is reportedly working with partners like MediaTek, Qualcomm, and Luxshare to develop an AI-native smartphone, with mass production targeted around 2028. The idea is simple but big: move away from app-based interfaces and toward a system where an AI agent handles tasks across the entire device, powered by a mix of on-device and cloud intelligence. By controlling both hardware and software, OpenAI could create a more seamless, context-aware experience than what’s possible on today’s smartphones. The bigger signal is where this is heading, AI companies aren’t just building apps anymore, they’re starting to rethink the device itself as the primary interface.
Microsoft researchers found that many of the roles most exposed to AI are traditional knowledge jobs like writers, analysts, translators, and even teachers, with tasks closely matching what AI can already do well. While that doesn’t mean these roles disappear, hiring is already slowing as companies rethink how much work needs to be done by humans. At the same time, lower-exposure roles tend to be physical or hands-on jobs that AI can’t easily replicate. The bigger signal is shifting expectations: a degree is no longer a guarantee of security, and the advantage is moving toward people who can work with AI rather than compete against it.
Chinese regulators have ordered Meta to reverse its $2.5 billion acquisition of Manus AI, escalating tensions around cross-border AI deals and signaling tighter control over strategic technologies. The case also cracks down on “Singapore washing,” where Chinese startups shift operations offshore to attract foreign investment. The bigger signal is geopolitical: AI is no longer just a competitive market, it’s becoming a regulated asset, and global expansion is starting to hit hard limits.
A Red Hat engineer has released Tank OS, a new tool designed to make OpenClaw deployments safer and easier to manage at scale by running agents inside isolated, containerized environments. The setup limits access between agents, protects credentials, and gives IT teams a way to manage fleets of autonomous systems using familiar infrastructure tools. The bigger signal is where this is heading: as AI agents move from experiments to enterprise workflows, security and control are quickly becoming the main bottlenecks, not capability.