Five items this morning, and four of them are really one item wearing different clothes: who gets to decide how fast AI moves, and on whose hardware. China’s AI Plan Now Runs on Huawei’s Delivery Calendar Liang Wenfeng rarely says anything in public, so the framing matters. Training DeepSeek’s models on Huawei silicon is, in his words, one of the company’s biggest bets. He called it a bet, which is a stronger word than a company hedging against tighter export controls would reach for.

The serious criticism of any voluntary international framework is that it has no teeth. No enforcement, no penalties, no court. Governments sign, praise it, and go home. The criticism is accurate and mostly beside the point, because binding treaties have become close to impossible to conclude among the states that matter, and the soft instruments that replaced them turn out to do a surprising amount of governing. Look at what’s happened to ratification.

G20 governments met in Chapel Hill this month and endorsed a set of AI policy principles backed by Washington. They’re non-binding. They favour rules written for particular uses over rules written for the technology itself, and they lean on cooperation between governments and industry. The name attached to them is the Carolina Principles. What they don’t do is create a universal regulatory regime, and that omission is the substance of the agreement.

Three regulatory actions landed inside twenty-four hours, in three jurisdictions, aimed at three different problems. What they have in common is that none of them used a statute written for the technology in question. The FCC’s ban on foreign advanced robotic devices covers any new ground-based, software-controlled wireless robot weighing more than 4.4 pounds. Because the threshold is defined by mass and locomotion rather than by capability, the rule reaches warehouse automation platforms, inspection robots, delivery machines, and consumer robot vacuums alike.

I think the security case about foreign-made connected devices is real. A camera-equipped, always-on machine that maps the interior of a home and phones a server abroad is a genuine collection platform, and the people who spent a decade warning about this were not being paranoid. I’ve written that a network of them inside government buildings would be an intelligence problem worth acting on. So I should be the natural audience for the FCC’s new rule.

Three regulatory outcomes in one day, all involving who is permitted to move goods and people through American infrastructure without a person in control. Waymo has begun gradually resuming freeway routes, starting in Phoenix, more than two months after suspending them to improve performance around construction zones. DoorDash received Part 135 air carrier certification from the FAA, allowing it to operate a drone delivery service, with a launch planned for the fall.

Integral Privacy Technologies has raised $25 million in total funding, with backing from Venrex, The General Partnership, Array Ventures, GreatPoint Ventures, LiveRamp Ventures, Haystack, Virtue Ventures, Also Capital, Caffeinated Capital, LifeX Ventures, Circle & Co, and WS Investments. The round is aimed at scaling what the company calls its Forward Deployed Privacy Services — a service layer meant to make sensitive real-world data usable for AI training without triggering the regulatory and re-identification risks that normally keep that data locked away.

For most of the history of industrial robotics, the machine and the task were synonymous. A welding robot welded. A pick-and-place robot picked and placed. The intelligence was in the fixture, the jig, the controlled environment — not in the robot. General purpose robotics represents the dismantling of that assumption, and the pace of that dismantling has accelerated sharply. The GAO’s 2026 horizon report frames the shift with the right analogy: robotics is following the trajectory of computers, from single-function devices to general platforms.