Amodei proposes agreed frontier pace, but who watches the watchers

Anthropic chief executive Dario Amodei published a plan this week for a controlled slowdown of frontier-model development, days after a researcher at the company warned of extreme-risk scenarios. The proposal rests on two pillars: independent safety evaluators who would vet capabilities before release, and coordination among AI labs in democratic countries. The idea has drawn some industry support, but also sharp criticism from Nvidia chief executive Jensen Huang, who argued such coordination could turn into a de facto cartel.

The missing mechanism

Amodei’s outline remains at the level of principle. There is no agreed definition of a “dangerous capability,” no roster of accredited evaluation bodies, and no enforcement mechanism for labs that decide to walk away from the consensus. Without binding quantitative benchmarks, each lab would be free to interpret the red lines as it sees fit. Huang zeroed in on precisely that gap: when the competition is for time-to-market and compute, the economic incentive to push ahead outweighs any voluntary commitment.

The coup at Automattic lasted 33 hours

Meanwhile, drama at WordPress parent Automattic: founder and chief executive Matt Mullenweg was ousted by the board and reinstated within 33 hours. During the brief interregnum, “golden parachute” agreements were signed with departing executives, a move that may ultimately strengthen Mullenweg’s grip on the technology powering a vast share of the web. Automattic’s hybrid structure — a private company intertwined with a major open-source project — turns any control battle into an issue affecting millions of sites.

SPAC and robotaxi: May Mobility bets on the public market

May Mobility, a developer of autonomous ride services, plans to go public via a SPAC merger that would net more than $300 million. The pitch: becoming a “public robotaxi company.” The reality: negligible revenue, reliance on partnerships with automakers, and regulation that varies by jurisdiction. SPACs in the autonomous space have proven adept at raising capital; returning it is another matter.

DoorDash invests in Wonder: automation from order to delivery

DoorDash will invest $425 million in Wonder, a startup building robotic kitchens and an integrated logistics network. The stated goal is full automation of the food chain, from order entry to the customer’s door. For DoorDash, the deal is a strategic hedge against rising labor costs; for Wonder, it funds expanded deployment. The open question is whether the technology has matured enough to replace human kitchens at economic scale, or whether this is a bet on a future that has not yet arrived.